Scribd Logo
Upload

Welcome to Scribd!

  • 133 views

    Midterm GR5412 2019 PDF

    Uploaded by

    tina
    This document appears to be an exam for an econometrics course covering topics like maximum likelihood estimation, Poisson regression, serial correlation, and time series analysis. It contains 3 questions asking students to define terms, derive aspects of Poisson regression like the log-likelihood function and MLE, and analyze a time series regression of vacancy and unemployment rates. Students are asked to consider issues like stationarity, exogeneity, serial correlation testing and correction, and normality of errors.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Midterm GR5412 2019 PDF

    Uploaded by

    tina
    133 views2 pages
    This document appears to be an exam for an econometrics course covering topics like maximum likelihood estimation, Poisson regression, serial correlation, and time series analysis. It contains 3 questions asking students to define terms, derive aspects of Poisson regression like the log-likelihood function and MLE, and analyze a time series regression of vacancy and unemployment rates. Students are asked to consider issues like stationarity, exogeneity, serial correlation testing and correction, and normality of errors.

    Original Title

    Midterm GR5412 2019.pdf

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document appears to be an exam for an econometrics course covering topics like maximum likelihood estimation, Poisson regression, serial correlation, and time series analysis. It contains 3 questions asking students to define terms, derive aspects of Poisson regression like the log-likelihood function and MLE, and analyze a time series regression of vacancy and unemployment rates. Students are asked to consider issues like stationarity, exogeneity, serial correlation testing and correction, and normality of errors.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    133 views2 pages

    Midterm GR5412 2019 PDF

    Uploaded by

    tina
    This document appears to be an exam for an econometrics course covering topics like maximum likelihood estimation, Poisson regression, serial correlation, and time series analysis. It contains 3 questions asking students to define terms, derive aspects of Poisson regression like the log-likelihood function and MLE, and analyze a time series regression of vacancy and unemployment rates. Students are asked to consider issues like stationarity, exogeneity, serial correlation testing and correction, and normality of errors.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Columbia

     University                                                                                              Econometrics  GR5412  


    Department  of  Economics     Spring  2019  
     
    Midterm  
     
    You  have  one  hour  and  fifteen  minutes.    Please  write  your  name  and  UNI  on  each  exam  
    booklet  that  you  use.  No  study  aids  are  allowed.  There  are  2  pages.  There  is  a  total  of  85  
    points.  
     
    1)  Definitions  (20  points  total)  
     
    Briefly  define  the  following  terms:  
    a) Posterior  mean                                          
    b) Truncated  observations  
    c) Davidson-­‐MacKinnon  test  
    d) Linear  index  model  for  binary  choice    
    e) Quantile  regression  (just  in  words  is  fine)  
     
    2)  (35  points  total)  Poisson  Regression:  Consider  a  variable  yi  that  only  takes  on  non-­‐
    negative  integer  values.    We  also  have  a  single  explanatory  variable,  xi.  Define  the  
    conditional  mean:  𝜇(𝑥! ) = 𝐸 𝑦! |𝑥!  and  assume  that  𝜇(𝑥) = 𝑒𝑥𝑝(𝛽𝑥).  Suppose  yi  varies  
    Poisson:    
    𝑓(𝑦|𝑥! ) = exp −𝜇 𝑥! 𝜇 𝑥! ! /𝑦!  
    a)  Derive  the  log-­‐likelihood  function  for  a  random  sample  of  size  n.    
    b)  Derive  the  score  and  show  that  the  expectation  of  the  score  at  the  true  value  of  β  is  
    zero.    
    c)  Describe  in  words  what  the  MLE  represents.    
    d)  Derive  the  Hessian.  
    e)  Describe  two  methods  by  which  you  could  numerically  calculate  the  MLE  for  a  specific  
    sample,  with  this  model.  
    f)  Consider  the  hypothesis  β=  β0.  Describe,  mathematically,  three  approaches  for  testing  
    this  hypothesis.  Briefly  discuss  their  strengths  and  weaknesses.  
     
     
     
     
     
     
     
     
     
     
     
     
     
     

        1  
       
     

    3)  (30  points  total)  You  are  estimating  a  Beveridge  curve  –  the  relationship  between  the  
    job  vacancy  rate  and  the  unemployment  rate  -­‐  with  monthly  time-­‐series  data.    You  
    estimate  a  time-­‐series  regression  of  the  job  vacancy  rate  on  the  unemployment  rate  and  
    a  linear  time  trend:  
    reg vrate urate trend

    Source | SS df MS Number of obs = 135


    -------------+---------------------------------- F(2, 132) = 209.98
    Model | 25.6891625 2 12.8445812 Prob > F = 0.0000
    Residual | 8.07454124 132 .061170767 R-squared = 0.7609
    -------------+---------------------------------- Adj R-squared = 0.7572
    Total | 33.7637037 134 .251967938 Root MSE = .24733

    ------------------------------------------------------------------------------
    vrate | Coef. Std. Err. t P>|t| [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    urate | -.241896 .013778 -17.56 0.000 -.2691504 -.2146417
    trend | .0041129 .0007845 5.24 0.000 .002561 .0056647
    _cons | 4.129086 .0675764 61.10 0.000 3.995413 4.262759
    ------------------------------------------------------------------------------

    a) Assuming  the  data  have  not  been  seasonally  adjusted,  if,  on  average  
    unemployment  rates  are  higher  in  the  summer,  are  the  data  covariance-­‐
    stationary?  Explain  briefly,  but  carefully.    
    b) Are  the  data  likely  to  be  strictly  exogenous?    Briefly  explain.    If  instead,  you  had  a  
    data  set  from  a  cross-­‐sectional  household  survey,  would  you  expect  that  data  to  
    be  strictly  exogenous?    Again,  explain.  
    c) Describe  precisely  how  you  would  construct  a  test  for  first  order  serial  
    correlation  in  this  model.  How  about  second  order  serial  correlation?  
    d) Suppose  your  test  finds  significant  serial  autocorrelation.    Are  the  OLS  estimates  
    above  consistent?    Unbiased?    Are  the  OLS  standard  errors  likely  to  be  correct?  
    e) Describe  exactly  how  you  could  use  OLS  regressions  to  estimate  a  FGLS  model  
    that  adjusts  for  first  order  serial  correlation.    Show  the  mathematics  and  show  
    that  the  approach  works.  
    f) Does  your  approach  in  (e)  require  additional  assumptions  in  order  to  guarantee  
    consistency,  as  compared  to  OLS.  
    g) The  dependent  variable  in  this  model  is  a  vacancy  rate.    As  such,  it  cannot  be  less  
    than  zero.    If  the  right  hand  side  of  the  regression  is  correctly  specified,  would  it  
    be  reasonable  to  assume  that  the  errors  are  normally  distributed?    Explain  
    briefly.  
     
     
     
     
     
     

      2  

    You might also like