%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Beamer Presentation
% LaTeX Template
% Version 1.0 (10/11/12)
%
% This template has been downloaded from:
% http://www.LaTeXTemplates.com
%
% License:
% CC BY-NC-SA 3.0 (http://creativecommons.org/licenses/by-nc-sa/3.0/)
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% ----------------------------------------------------------------------------------------
%	PACKAGES AND THEMES
% ----------------------------------------------------------------------------------------

\documentclass[hyperref={colorlinks=true}]{beamer}

\mode<presentation> {

  % The Beamer class comes with a number of default slide themes
  % which change the colors and layouts of slides. Below this is a list
  % of all the themes, uncomment each in turn to see what they look like.

  \usetheme{default}
  % \usetheme{AnnArbor}
  % \usetheme{Antibes}
  % \usetheme{Bergen}
  % \usetheme{Berkeley}
  % \usetheme{Berlin}
  % \usetheme{Boadilla}
  % \usetheme{CambridgeUS}
  % \usetheme{Copenhagen}
  % \usetheme{Darmstadt}
  % \usetheme{Dresden}
  % \usetheme{Frankfurt}
  % \usetheme{Goettingen}
  % \usetheme{Hannover}
  % \usetheme{Ilmenau}
  % \usetheme{JuanLesPins}
  % \usetheme{Luebeck}
  % \usetheme{Madrid}
  % \usetheme{Malmoe}
  % \usetheme{Marburg}
  % \usetheme{Montpellier}
  % \usetheme{PaloAlto}
  % \usetheme{Pittsburgh}
  % \usetheme{Rochester}
  % \usetheme{Singapore}
  % \usetheme{Szeged}
  % \usetheme{Warsaw}

  % As well as themes, the Beamer class has a number of color themes
  % for any slide theme. Uncomment each of these in turn to see how it
  % changes the colors of your current slide theme.

  % \usecolortheme{albatross}
  % \usecolortheme{beaver}
  % \usecolortheme{beetle}
  % \usecolortheme{crane}
  % \usecolortheme{dolphin}
  % \usecolortheme{dove}
  % \usecolortheme{fly}
  % \usecolortheme{lily}
  % \usecolortheme{orchid}
  \usecolortheme{rose}
  % \usecolortheme{seagull}
  % \usecolortheme{seahorse}
  % \usecolortheme{whale}
  % \usecolortheme{wolverine}

  % \setbeamertemplate{footline} % To remove the footer line in all slides uncomment this line
  % \setbeamertemplate{footline}[page number] % To replace the footer line in all slides with a simple slide count uncomment this line

  % \setbeamertemplate{navigation symbols}{} % To remove the navigation symbols from the bottom of all slides uncomment this line
}

\usepackage{graphicx} % Allows including images
\usepackage{booktabs} % Allows the use of \toprule, \midrule and \bottomrule in tables
\usepackage{cancel}
\usepackage{amsmath}
\usepackage{amssymb}
% \usepackage{showframe}
\usepackage{caption}

%\usepackage{subcaption}
\usepackage{tcolorbox}
\usepackage{tabularx}
\usepackage{array}
\usepackage{pgfplots}
\tcbuselibrary{skins}


\usepackage{subfig}
\beamertemplatenavigationsymbolsempty
\usepackage{color, colortbl}
\definecolor{LRed}{rgb}{1,.8,.8}
\definecolor{MRed}{rgb}{1,.6,.6}

\usepackage{tikz}
\usetikzlibrary{shapes,arrows,shapes.multipart,fit,shapes.misc,positioning}
\newcommand{\der}[2]{\frac{\partial #1}{\partial #2}}
%\usepackage[labelformat=empty]{caption}

\newcommand{\ra}[1]{\renewcommand{\arraystretch}{#1}}

\newcommand\marktopleft[1]{%
    \tikz[overlay,remember picture]
        \node (marker-#1-a) at (0,1.5ex) {};%
}
\newcommand\markbottomright[1]{%
    \tikz[overlay,remember picture]
        \node (marker-#1-b) at (7ex,0) {};%
    \tikz[overlay,remember picture,thick,dashed,inner sep=3pt]
        \node[draw=black,rounded corners=0pt,fill=red,opacity=.2,fit=(marker-#1-a.center) (marker-#1-b.center)] {};%
}


% ----------------------------------------------------------------------------------------
%	TITLE PAGE
% ----------------------------------------------------------------------------------------

\title[]{Quantitative Macro-Labor\\ Empirical Regularities and Panel Data} % The short title appears at the bottom of every slide, the full
% title is only on the title page

\author{Professor Griffy} % Your name
\institute[University at Albany, SUNY] % Your institution as it will appear on the bottom of
% every slide, may be shorthand to save space
{
UAlbany  \ % Your institution for the title page
}
\date{Fall 2024} % Date, can be changed to a custom date

\begin{document}

\begin{frame}
  \titlepage % Print the title page as the first slide
\end{frame}




% ----------------------------------------------------------------------------------------
%	PRESENTATION SLIDES
% ----------------------------------------------------------------------------------------

% ------------------------------------------------
\section{Course Introduction} % Sections can be created in order to organize your presentation into discrete blocks, all sections and subsections are automatically printed in the table of contents as an overview of the talk
% ------------------------------------------------

\begin{frame}
  \frametitle{Announcements}
  \begin{itemize}
  \item I still need to email about the campus cluster--moving was chaos.
  \item You can run Matlab/Python/R/Julia code on the campus cluster, and it will email you when done.
  \item Stata should work as well, let me know if you want to use it.
  \end{itemize}
\end{frame}

% ------------------------------------------------
\section{Panel Data} % Sections can be created in order to organize your presentation into discrete blocks, all sections and subsections are automatically printed in the table of contents as an overview of the talk
% ------------------------------------------------

\begin{frame}
  \frametitle{Panel Data}
  \begin{itemize}
  \item What is panel data?
    \begin{enumerate}
    \item Repeated surveys of the same individuals.
    \item Surveys contain repeated questions, thus comparable across time/age.
    \item Generally, introduction of {\it new} cohorts, thus allowing time and age effects to be disentangled.
    \end{enumerate}
  \item Why is it useful?
    \begin{enumerate}
    \item Repeated individual observations help separate marginal effects of observables from innate ability.
    \item (With enough data), individual fixed effects control for time-invariant innate characteristics.
    \item Can control for geography-by-time trends, as well as the marginal effects of other ``nuissance'' covariates.
    \end{enumerate}
  \item Excellent discussion of use in macroeconomics: Browning, Heckman, and Hansen (1999).
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{``Fundamental Equation of (Reduced-Form) Labor Economics''}
  \begin{itemize}
  \item Standard regression analysis:
    \begin{align}
      y &= X\beta + \epsilon
    \end{align}
  \item Most of ``reduced-form'' labor economics comes down to arguing the following:
    \begin{align}
      E[X\epsilon] &= 0\\
      \text{or   } E[\epsilon|X] &= 0
    \end{align}
  \item i.e., that your covariates are uncorrelated with the error term,
  \item or alternatively that you aren't capturing variation with a covariate that is actually caused by an omitted variable.
  \item If you can successfully argue this, you have argued for {\it ex-post} identification.
  \item Note: ``reduced-form'' is not intended as a pejorative.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Two Basic Panel Models}
  \begin{itemize}
  \item Fixed Effects:
    \begin{itemize}
    \item ``Fixed Effect Model'' means that you have {\it individual} (or firm, etc.) fixed effects in your regression.
    \item i.e., an intercept for every individual.
    \item Don't confuse this with {\it using} fixed effects, i.e., state, year.
    \end{itemize}
  \item Random Effects:
    \begin{itemize}
    \item There is an individual unobserved heterogeneity, but it is {\it random}, i.e., uncorrelated with your observable characteristics.
    \item I can't come up with a good example of this, and in almost every case people use fixed effects models.
    \end{itemize}
  \item We'll focus briefly on the fixed effects model.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Fixed Effects Model}
  \begin{itemize}
  \item Generic linear regression:
    \begin{align}
      y_{it} &= x_{it}\beta + c_{i} + \epsilon_{it}
    \end{align}
  \item $c_{i}$ is the individual heterogeneity/effect.
  \item Typically, we would run
    \begin{align}
            y_{it} &= x_{it}\beta + \epsilon_{it}
    \end{align}
  \item But, this would be wrong if $E[X_{it}c_{i}] \cancel{=} 0$.
  \item ``The point of using panel data is to allow $c_{i}$ to be arbitrarily correlated with the $x_{it}$'' (Wooldridge, 2002).
  \item Some good references:
    \begin{itemize}
    \item ``Econometric Analysis of Cross Section and Panel Data'': Wooldridge (2002)
    \item ``Mostly Harmless Econometrics'': Angrist and Pischke (2009)
    \end{itemize}
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Fixed Effects Model}
  \begin{itemize}
  \item Generic linear regression:
    \begin{align}
      y_{it} &= x_{it}\beta + c_{i} + \epsilon_{it}
    \end{align}
  \item How do we solve this problem?
    \begin{itemize}
    \item ``within transformation'' FE estimator
    \item ``first-difference'' estimator
    \end{itemize}
  \item Within-transformation: difference out the mean over time of each observation
    \begin{align}
      y_{it} - \bar{y}_{i} &= x_{it}\beta + c_{i} + \epsilon_{it} - \bar{y}_{i}\\
      y_{it} - \bar{y}_{i} &= x_{it}\beta + c_{i} + \epsilon_{it} - \bar{x}_{i}\beta - \bar{c}_{i} - \bar{\epsilon}_{i}\\
      y_{it} - \bar{y}_{i} &= (x_{it} - \bar{x}_{i})\beta + \cancel{(c_{i} - \bar{c}_{i})} + (\epsilon_{it} - \bar{\epsilon}_{i})\\
      y_{it} - \bar{y}_{i} &= (x_{it} - \bar{x}_{i})\beta + (\epsilon_{it} - \bar{\epsilon}_{i})\\
    \end{align}
  \item Identical to having an indicator variable for each individual.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{First-Difference Approach}
  \begin{itemize}
  \item Generic linear regression:
    \begin{align}
      y_{it} &= x_{it}\beta + c_{i} + \epsilon_{it}
    \end{align}
  \item First-difference estimator:
    \begin{align}
      y_{it} - y_{it-1} &= x_{it}\beta + c_{i} + \epsilon_{it} - y_{it-1}\\
      y_{it} - y_{it-1} &= x_{it}\beta + c_{i} + \epsilon_{it} - x_{it-1}\beta - c_{i} - \epsilon_{it-1}\\
      y_{it} - y_{it-1} &= (x_{it}-x_{it-1})\beta + \cancel{(c_{i} - c_{i})} + (\epsilon_{it}-\epsilon_{it-1})\\
      y_{it} - y_{it-1} &= (x_{it}-x_{it-1})\beta + (\epsilon_{it}-\epsilon_{it-1})
    \end{align}
  \item Identical to fixed-effects estimator if errors not serially correlated.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Guvenen (2009)}
  \begin{itemize}
  \item What Guvenen ultimately ends up estimating:
    \begin{align}
      y_{h,t}^{i} &= g(\theta_{t}^{0},X_{h,t}^{i}) + c^{i} + d^{i}\times t + z_{h,t}^{i} + \phi_{t}\epsilon_{h,t}^{i}\\
      y_{h,t}^{i} &= g(\theta_{t}^{0},X_{h,t}^{i}) + c^{i} + d^{i}\times t + \hat{\epsilon}_{h,t}^{i}\\
      c^{i}&: Ind.\;FE\;that\;affects\;intercept\\
      d^{i}&: Ind.\;FE\;that\;affects\;slope
    \end{align}
  \item where $\hat{\epsilon}_{h,t}^{i}$ includes all unobserved components (persistent and transitory shocks).
  \item Estimate this jointly with transition equation for $z_{it}$ to recover $\rho$, $\sigma_{\epsilon}$ and $\sigma_{\nu}$.
  \end{itemize}
\end{frame}

% ------------------------------------------------


\begin{frame}
  \frametitle{Fixed Effects Model}
  \begin{itemize}
  \item What does this mean?
    \begin{itemize}
    \item If unobserved heterogeneity is {\bf not} time-varying,
    \item and we have correctly specified our model,
    \item we can identify the marginal effect, $\beta$, of each covariate in $x_{it}$.
    \end{itemize}
  \item We are using ``within individual'' variation to identify the effects.
  \item Potential problems:
    \begin{itemize}
    \item The covariates of interest may also be time-invariant.
    \item May have relatively few individual-level observations.
    \item Then, we would use between-individual variation and try to argue that our inference can be interpreted causally.
    \item Or use a structural model to try and interpret our results.
    \end{itemize}
  \end{itemize}
\end{frame}

% ------------------------------------------------

\section{Sources of Panel Data} % Sections can be created in order to organize your presentation into discrete blocks, all sections and subsections are automatically printed in the table of contents as an overview of the talk
% ------------------------------------------------

\begin{frame}
  \frametitle{Some Valuable Micro-Data Sources}
  \begin{itemize}
  \item In class, we will typically discuss ``micro-data'':
    \begin{enumerate}
    \item Panel Study of Income Dynamics (PSID): a panel of households from 1968-present, annually.
    \item National Longitudinal Survey of Youth (NLSY79, NLSY97): Two separate cohorts interviewed repeatedly at an annual frequency 1979-present and 1997-present.
    \item Survey of Income and Program Participation (SIPP): Series of panels that last 3-4 years. Each panel contains new participants.
    \item Current Population Survey (CPS): The standard for labor market information. A monthly survey that is representative. Some panel dimensions, but note that this lacks important panel components that the others retain.
    \end{enumerate}
  \item I will upload some code to the lab storage.
  \item Link to a good description on website.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Panel Study of Income Dynamics}
  \begin{itemize}
  \item Longitudinal study of a representative sample of US individuals and their families from 1968-present.
  \item New individuals enter and exit, meaning many cohorts.
  \item Excellent panel for life-cycle analysis (almost the exclusive source of data).
  \item Good labor market information: employment spells, income, wages, some employer-to-employer and job-to-job mobility.
  \item The bad:
    \begin{itemize}
    \item Annual frequency.
    \item (potentially) substantial measurement error.
    \item Can be hard to work with: variables renamed each year.
    \end{itemize}
  \item Can be used for intergenerational analysis as well (only dataset that can).
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Survey of Income and Program Participation}
  \begin{itemize}
  \item The SIPP is a series of short panels, rarely more than 3 years in length.
  \item Conducted annually 1984-1993, then in 1996, 2001, 2004, 2008.
  \item Households are assigned a ``rotation group,'' and interviewed every four months about the previous four months.
  \item Great for labor market information: weekly labor force status, income, hours, wages, UI, mobility, etc.
  \item The bad:
    \begin{itemize}
    \item It's a very short panel: no life-cycle components
    \item Might only observe a single unemployment spell by individuals
    \item Problems with censoring because of survey length
    \item Survey design is a little tricky
    \end{itemize}
  \item Probably best publicly available panel data for labor market.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{National Longitudinal Survey of Youth (1979)}
  \begin{itemize}
  \item The NLSY is (sort-of) a medium between the PSID and the SIPP.
  \item A cohort of 14-22 year olds are surveyed identical questions each year from 1979 to present.
  \item Has very detailed labor market information, and can be at a monthly frequency.
  \item Best (IMHO): has Armed Forces Qualifying Test (AFQT) scores, which are a rough measure of individual ability.
  \item Also has relatively consistent wealth observations.
  \item The bad:
    \begin{enumerate}
    \item Annual frequency;
    \item Single cohort;
    \item Geographic information only available in restricted version.
    \end{enumerate}
  \item Easiest of the 3 to work with.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Micro-Data}
  \begin{itemize}
  \item Which should you work with? Depends on the question.
  \item Broadly,
    \begin{enumerate}
    \item If you aren't interested in life-cycle effects, choose the SIPP.
    \item If you are, need geographic location, or need to separate time and age effects, choose the PSID.
    \item If you want a measure of individual ability, the AFQT, choose the NLSY.
    \end{enumerate}
  \item If you aren't interested in the panel dimension, choose the CPS.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\section{Labor Market Empirical Regularities} % Sections can be created in order to organize your presentation into discrete blocks, all sections and subsections are automatically printed in the table of contents as an overview of the talk
% ------------------------------------------------

\begin{frame}
  \frametitle{Labor Market Empirical Regularities}
  \begin{itemize}
  \item What are some topics that are worth exploring in the labor market?
  \item Davis and Haltiwanger (1999) talk about six:
    \begin{enumerate}
    \item Employer lifecycle dynamics;
    \item Worker reallocation and productivity growth;
    \item Worker reallocation over business cycles;
    \item Lumpiness, heterogeneity, and aggregation;
    \item Reasons for worker mobility;
    \item Worker sorting and job assignment.
    \end{enumerate}
  \item They argue that each of these topics (at least at the time) had unanswered questions.
  \item Davis and Haltiwanger papers (there are a lot) are a good source of ``empirical regularities''
  \item FYI: they also use the term ``empirical regularities''
  \end{itemize}
\end{frame}

% ------------------------------------------------


\begin{frame}
  \frametitle{Davis and Haltiwanger (1999)}
  \begin{itemize}
  \item Lots of interest in worker flows.
  \item They are purely interested in measurement.
  \item Challenges:
    \begin{enumerate}
    \item Few matched employer-employee data sets.
    \item Aggregation issues: flows between plants within same firm, etc.
    \item Few matched employer-employee panels, i.e., can't separate worker and firm fixed effects.
    \end{enumerate}
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Davis and Haltiwanger (1999) - Key Definitions}
  \begin{itemize}
  \item Primary Definitions:
    \begin{itemize}
    \item (Gross) job creation at time $t$ equals employment gains summed over all business units that expand or start up between $t - 1$ and $t$.
    \item (Gross) job destruction at time $t$ equals employment losses summed over all business units that contract or shut down between $t - 1$ and $t$.
    \end{itemize}
  \item Secondary Definitions:
    \begin{itemize}
    \item (Gross) job reallocation: job creation $+$ job destruction
    \item Excess job reallocation: job reallocation $-$ net employment change
    \item (Gross) {\bf worker} reallocation: movement across place of employment.
    \end{itemize}
  \item Excess reallocation: the amount of job reallocations {\it over and above} the amount required to accomodate net employment changes.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Measurement and Job/Worker Flows}
  \begin{itemize}
  \item Tricky Proposition:
    \begin{itemize}
    \item Flow is necessarily a continuous variable;
    \item Surveys yield snapshots at various points in time.
    \item Workers may transition jobs before being observed, i.e., $E_{t}, U_{t+0.5\times\Delta}, E_{t + \Delta}$.
    \end{itemize}
  \item Definitions change over time.
  \item Workers flow E-O and O-E, so even out of labor force are searching for jobs.
  \item Rob Shimer has a lot of good work on this as well.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Previous work across countries}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger1.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Between vs. within employers}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger2.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{By industry}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger3.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Excess Reallocation}
\centering\includegraphics[width=0.7\textwidth]{Davis_Haltiwanger4.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Persistence}
  \begin{itemize}
  \item Persistence of job creation: \% of jobs at time $t$ that remain filled at $t + n$
  \item Persistence of job destruction: \% of jobs at time $t$ that do not reappear by $t + n$
  \end{itemize}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger5.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger6.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger7.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Size}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger8.png}
\begin{itemize}
\item Declines with age.
\end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Size}
\centering\includegraphics[width=0.9\textwidth]{Davis_Haltiwanger9.png}
\begin{itemize}
  \item Excess reallocation: the amount of job reallocations {\it over and above} the amount required to accomodate net employment $\Delta$.
  \item Declines with age.
\end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Business Cycles}
\centering\includegraphics[width=\textwidth]{Davis_Haltiwanger10.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{A Follow-Up: Davis, Faberman, Haltiwanger (2006)}
  \begin{itemize}
  \item Again on measurement (they've done a lot of good work on it).
  \item Comparison of results across different datasets.
    \begin{itemize}
    \item JOLTS: Job Openings and Labor Turnover Survey
    \item BED: Business Employment Dynamics
    \item LEHD: Longitudinal Employer Household Dynamics
    \end{itemize}
  \item Updated with new papers and findings.
  \end{itemize}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{DavisEtAl1.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{DavisEtAl2.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{DavisEtAl3.png}
\end{frame}

% ------------------------------------------------

\begin{frame}
  \frametitle{Firm Distribution of Growth Rates}
\centering\includegraphics[width=0.9\textwidth]{DavisEtAl4.png}
\end{frame}

% ------------------------------------------------

\section{Conclusion}
% ------------------------------------------------


\begin{frame}
  \frametitle{Next Time}
  \begin{itemize}
  \item Job search: how can we explain wage dispersion?
  \item The McCall Model.
  \item Read Rogerson, Shimer, Wright (2005).
  \item Make sure you've installed some programming languages.
  \end{itemize}
\end{frame}


\end{document}