Untitled Flashcards Set

Spatial Weighting Matrices

Tools used to capture spatial relationships between observations, helping to address omitted variable bias and improve the accuracy of standard error estimates.

Bias in Spatial Econometrics

Occurs when omitted spatial factors influence the outcome variable; can be addressed using spatial weighting matrices.

Standard Errors in Spatial Econometrics

Traditional standard errors can be underestimated due to spatial correlation; spatial weighting matrices help construct robust standard errors.

Cutoff Distance for Conley Standard Errors

The distance used in spatial econometrics to determine the range of spatial correlation; affects computational efficiency and accuracy in correcting for bias.

Reflection Problem

A challenge in spatial models where endogenous effects cannot be distinguished from contextual effects due to collinearity.

LASSO Method

A statistical method for linear regression that includes automatic feature selection and is useful for policy impact studies.

Random Forest Method

A machine learning technique that captures complex nonlinear relationships and interactions, often used for prediction tasks.

SUTVA

Stable Unit Treatment Value Assumption, which posits that the treatment of one unit does not affect other units; violated by spatial spillovers.

Spatial Regression Discontinuity Design

A method of causal inference that can be complicated by overlapping administrative and treatment boundaries, affecting the continuity assumption.

Cross Validation

A method for assessing a model's ability to generalize by partitioning data into training and validation sets.

Google Earth Engine

A cloud-based platform that facilitates large-scale processing and analysis of satellite data.

Spatial Data Formats

Different types of spatial data structures, including raster, polygon, and point shapefiles, used in various economic applications.

Machine Learning in Control Variable Selection

Uses automated methods to select control variables when theoretical guidance is limited, improving model efficiency.

Sources of Bias in Spatial Models

Include omitted spatial variables and spatial spillovers, which can be addressed with fixed effects, spatial differencing, and careful control group selection.

Data Construction for Spatial RD Design

Involves obtaining boundary data, assigning treatment status, and ensuring accurate calculations and appropriate bandwidth.

Spatial Autocorrelation

Occurs when residuals from observed data are correlated due to geographic proximity, potentially inflating statistical significance.

Satellite Data in Economic Research

Provides objective measurements of economic activity and can be used to study areas lacking traditional data.

Fixed Effects vs. Spatial Differencing

Fixed effects control for time-invariant factors; spatial differencing addresses spatially varying unobservables, each with different applicability.

Cluster Randomization Challenges

Includes issues with reflection problems, intracluster correlation, and the need for additional control measures in spatial studies.

Testing for Nonrandomness in Spatial Data

Utilizes various statistical methods like point pattern analysis and indicators of spatial association to assess spatial randomness.

Endogenous vs. Contextual Effects

Endogenous effects relate to how peer outcomes influence individuals, while contextual effects relate to the influence of peer characteristics.

Criteria for Choosing LASSO vs. Random Forest

Factors such as linearity, interpretability, computational intensity, and presence of interactions guide the choice between both methods.

Modifiable Areal Unit Problem (MAUP)

A phenomenon where estimates vary significantly with different spatial aggregations, exacerbated by spatial correlation.

Specifying a Spatial Weights Matrix

Involves consideration of structure, normalization, and treatment of self-connections, impacting model interpretation and identification.

Addressing Weak Instruments in Spatial Models

Strategies include using higher-order spatial lags or features providing exogenous variation in the context of instrumental variable estimation.

Spatial Correlation and Treatment Effects

Spatial correlation affects the precision of treatment effect estimates, requiring adjustments in standard errors and careful model selection.