Python API

Overview

The main PyTwoWay API is split into multiple classes for estimating models and two for simulating data according to the BLM and Borovickova-Shimer dgps. It also has four modules: one for constructing the variance-covariance matrix for AKM and its bias corrections; one for generating constraints for BLM; one for generating attrition plots using increasing (building up from a fixed set of firms) or decreasing (with varying sets of firms) fractions of movers; and one for generating diagnostics for bipartite labor data. PyTwoWay is canonically imported using

import pytwoway as tw

Classes

• pytwoway.FEEstimator: Class for estimating AKM and its bias corrections without controls

• pytwoway.FEControlEstimator: Class for estimating AKM and its bias corrections with controls

• pytwoway.CREEstimator: Class for estimating the CRE model

• pytwoway.BLMModel: Class for estimating the BLM model with one set of starting values

• pytwoway.BLMEstimator: Class for estimating the BLM model with multiple sets of starting values

• pytwoway.BLMBootstrap: Class for estimating the BLM model with bootstrapped confidence intervals

• pytwoway.BLMVarianceDecomposition: Class for estimating the variance decomposition of the BLM model with bootstrapped confidence intervals

• pytwoway.InteractedBLMModel: Class for estimating the interacted BLM model

• pytwoway.SorkinEstimator: Class for estimating the fixed-point revealed preference model from Sorkin

• pytwoway.BSEstimator: Class for estimating the non-parametric sorting model from Borovickova and Shimer

• pytwoway.Attrition: Class for generating attrition plots

• pytwoway.MonteCarlo: Class for running Monte Carlo estimations

• pytwoway.SimBLM: Class for simulating BLM data

• pytwoway.SimBS: Class for simulating Borovickova-Shimer data

Modules

• pytwoway.Q: Module for constructing the variance-covariance matrix for AKM and its bias corrections

• pytwoway.constraints: Module for generating constraints for BLM

• pytwoway.attrition_utils: Module for estimating attrition using increasing (building up from a fixed set of firms) or decreasing (with varying sets of firms) fractions of movers

• pytwoway.diagnostics: Module for generating diagnostics for bipartite labor data

Classes and Methods

pytwoway.FEEstimator

FEEstimator(adata[, params])	Uses multigrid and partialing out to solve two way fixed effect models.
fit([rng])	Estimate FE model.

pytwoway.FEControlEstimator

FEControlEstimator(data[, params])	Solve two way fixed effect models with control variables.
fit([rng])	Estimate FE model.

pytwoway.CREEstimator

CREEstimator(data[, params])	Uses multigrid and partialing out to solve two way Fixed Effect model.
fit([rng])	Run CRE solver.

pytwoway.BLMModel

BLMModel(params[, rng])	Class for estimating BLM using a single set of starting values.
compute_connectedness_measure([all])	Computes graph connectedness measure among the movers within each type and updates self.connectedness to be the smallest value.
fit_movers(jdata[, compute_NNm])	EM algorithm for movers.
fit_movers_cstr_uncstr(jdata[, compute_NNm])	Run fit_movers(), first constrained, then using results as starting values, run unconstrained.
fit_stayers(sdata[, compute_NNs])	EM algorithm for stayers.
plot_log_earnings([period, grid, dpi])	Plot log-earnings by worker-firm type pairs.
plot_type_proportions([period, subset, dpi])	Plot proportions of worker types at each firm class.

pytwoway.BLMEstimator

BLMEstimator(params)	Class for estimating BLM using multiple sets of starting values.
fit(jdata, sdata[, n_init, n_best, ncore, rng])	Estimate BLM using multiple sets of starting values.
plot_liks_connectedness([jitter, dpi])	Plot likelihoods vs connectedness for the estimations run.
plot_log_earnings([period, grid, dpi])	Plot log-earnings by worker-firm type pairs.
plot_type_proportions([period, subset, dpi])	Plot proportions of worker types at each firm class.

pytwoway.BLMBootstrap

BLMBootstrap(params)	Class for estimating BLM using bootstrapping.
fit(jdata, sdata[, blm_model, n_samples, ...])	Estimate bootstrap.
plot_log_earnings([period, grid, dpi])	Plot log-earnings by worker-firm type pairs.
plot_type_proportions([period, subset, dpi])	Plot proportions of worker types at each firm class.

pytwoway.BLMVarianceDecomposition

BLMVarianceDecomposition(params)	Class for estimating BLM variance decomposition using bootstrapping.
fit(jdata, sdata[, blm_model, n_samples, ...])	Estimate variance decomposition.

pytwoway.InteractedBLMModel

pytwoway.SorkinEstimator

SorkinEstimator()	Class for estimating the fixed-point revealed preference model from Sorkin (2018).
fit(adata[, max_iters, threshold])	Estimate the fixed-point revealed preference model from Sorkin (2018).

pytwoway.BSEstimator

BSEstimator()	Class for estimating the non-parametric sorting model from Borovickova and Shimer.
fit(adata[, alternative_estimator, weighted])	Estimate the non-parametric sorting model from Borovickova and Shimer.

pytwoway.Attrition

Attrition([min_movers_threshold, ...])	Class of Attrition, which generates attrition plots using bipartite labor data.
attrition(bdf[, N, ncore, copy, rng])	Run Monte Carlo on attrition estimations of TwoWay to estimate variance of parameter estimates given fraction of movers remaining.
boxplots([fe, ho, he, cre, bs1, bs2, ...])	Generate attrition result boxplots.
plots([fe, ho, he, cre, bs1, bs2, ...])	Generate attrition result plots.

pytwoway.MonteCarlo

MonteCarlo([sim_params, fe_params, ...])	Class of MonteCarlo, where MonteCarlo runs a Monte Carlo estimation by simulating bipartite networks of firms and workers.
hist([fe, ho, he, cre, bs1, bs2, density])	Plot histogram of how Monte Carlo simulation results differ from truth.
monte_carlo([N, ncore, rng])	Run Monte Carlo simulations of two way fixed effect models to see the distribution of the true vs.

pytwoway.SimBLM

SimBLM([sim_params])	Class of SimBLM, where SimBLM simulates a bipartite BLM network of firms and workers.
simulate([return_parameters, rng])	Simulate data (movers and stayers).

pytwoway.SimBS

SimBS([sim_params])	Class of SimBS, where SimBS simulates a bipartite Borovickova-Shimer network of firms and workers.
simulate([rng])	Simulate data.

Modules and Methods

pytwoway.Q

CovCovariate(cov_names_1, cov_names_2)	Generate Q to estimate cov(covariate 1, covariate 2).
CovPsiAlpha()	Generate Q to estimate cov(psi, alpha).
CovPsiPrevPsiNext()	Generate Q to estimate cov(psi_t, psi_{t+1}).
VarAlpha()	Generate Q to estimate var(alpha).
VarCovariate(cov_names)	Generate Q to estimate var(covariate) or var(sum(covariates)) if multiple covariates are listed.
VarPsi()	Generate Q to estimate var(psi).
VarPsiPlusAlpha()	Generate Q to estimate var(psi + alpha).

pytwoway.constraints

BoundedAbove([ub, nt])	Generate BLM constraints so that worker-firm pair effects are bounded above.
BoundedBelow([lb, nt])	Generate BLM constraints so that worker-firm pair effects are bounded below.
Linear([nnt, nt])	Generate BLM constraints so that for a fixed firm type, worker types effects must change linearly.
Monotonic([md, increasing, nt])	Generate BLM constraints so that for a fixed firm type, worker types effects must increase (or decrease) monotonically.
NoWorkerTypeInteraction([nnt, nt])	Generate BLM constraints so that for a fixed firm type, worker types effects must all be the same.
QPConstrained(nl, nk)	Solve a quadratic programming model of the following form:
Stationary([nwt, nt])	Generate BLM constraints so that worker-firm pair effects are the same in all periods.
StationaryFirmTypeVariation([nnt, nt])	Generate BLM constraints so that the firm type induced variation of worker-firm pair effects is the same in all periods.

pytwoway.attrition_utils

AttritionIncreasing([subset_fractions, ...])	Generate increasing subsets of a dataset to estimate the effects of attrition.
AttritionDecreasing([subset_fractions])	Generate decreasing subsets of a dataset to estimate the effects of attrition.

pytwoway.diagnostics

plot_extendedeventstudy(adata[, ...])

Generate event study plots.