||Macroeconomic analysis studies the relationship between economic variables that are measured at an aggregate level: GDP, growth, unemployment, inflation. It often distinguishes between short-term effects, i.e. variation within the business cycle, and long-term effects that generate economic growth.
||Micro-level analysis is concerned with the behavior of individuals or firms, and, in particular, the decision to supply or the demand for goods or services. The analysis can be conducted under perfect competition or with some forms of market failure. Microeconomics can be understood as the study of the underlying mechanisms behind the macro-level analysis.
||Descriptive statistics summarize the basic characteristics of variables, often reporting the mean value and a measure of dispersion of the observations or a graph of the variable distribution.
||A correlation measures the degree by which two or more variables’ movements are dependent. The dependence can be positive, i.e. the two variables move in the same direction, or negative. Correlation coefficients range from -1 to 1. A coefficient of unity indicates that the two variables are perfectly (linearly) related. It is important to note that correlation does not imply causation (see Causality).
||A relationship is causal if a change in variable x (the cause) leads to an adjustment in variable y (the effect). Estimating causality involves defining a counter-factual event: what would have been the value of y if x had remained at its initial value. This is never observed, and needs to be estimated. Econometric methods rely on different assumptions for estimating the counter-factual value. The size of the effect is then the difference between the observed value of y and the counter-factual value.
||Regression is a statistical tool used to establish how much a dependent variable varies when another variable varies. Compared to correlation, regression allows for the controlling of a set of additional variables that may affect the relationship of interest. Several regression techniques exist, imposing various assumptions on the relationship of interest. It is important to note again that a statistically significant regression coefficient does not imply a causal relationship between the independent variables and the dependent variable, but depending on the hypothesis written, causality may be inferred (see Identification strategy).
|| This is the strategy adopted in order to causally interpret the result of a regression. The basic idea of an identification strategy is for hypotheses to be written in such a way that the data become as close as possible to a randomized control trial; i.e. whereby the individuals affected by a policy are randomly selected and thus identical to those not receiving it. The latter group is termed the control group and provides the counter-factual outcome (see Causality) of what would have happened in the absence of the policy.
||A meta study statistically analyses results from previous studies on a given topic to identify common patterns in their results and to provide a summary of current knowledge.
||Simulation is a statistical practice that involves modeling a process using a set of equations; for example, a model of unemployment would include the job search effort of individuals, as well as the demand for labor from firms. Once this model has been calibrated—so that it can replicate the observed relationship—it can then be used to simulate the potential effect of policies on the outcome of interest.