In asymptotic theory, we deal with sequences of random variables. The concept of a limit point (in probability, almost surely) comes directly from the topological foundations of analysis.
Most economics students encounter calculus —rules for differentiation, integration, and optimization. Few, however, are introduced to mathematical analysis : the rigorous foundation upon which calculus rests. At first glance, this seems reasonable. Why spend weeks on epsilon-delta proofs, compactness, or measure theory when you need to estimate a demand curve or solve a dynamic programming problem?
To prove consistency and rates of convergence, we use tools from : entropy, covering numbers, and Donsker classes—all built on measure theory and functional analysis.
: Unlike many texts that treat mathematical topics in isolation, this book uses a unified approach centered on the Metric Completion Theorem In asymptotic theory, we deal with sequences of
If you pick two points in a set, the line connecting them is also in the set.
Part II of the book (chapters 8–10) focuses on without full measure theory overload. You will learn why “almost sure convergence” is stronger than convergence in probability, why continuous mapping theorem holds, and how to prove consistency of an extremum estimator (e.g., MLE, GMM) using uniform laws of large numbers—without feeling lost.
For econometrics, move to:
In nonlinear econometrics (e.g., maximum likelihood, GMM), we maximize a sample objective function ( Q_n(\theta) ) that converges pointwise to ( Q(\theta) ). To guarantee that the maximizer ( \hat\theta n ) converges to ( \theta_0 ), we need : [ \sup \theta \in \Theta |Q_n(\theta) - Q(\theta)| \to 0 \quad \textin probability ]
: Crucial for proving the existence of market equilibrium and stable strategic outcomes in game theory.
To understand the applications of mathematical analysis in economics, it is essential to grasp some basic concepts, including: Few, however, are introduced to mathematical analysis :
Uses analysis to show that the distribution of many independent variables tends toward a normal distribution.
Instead, we follow a targeted path: