T 5 s a t ² u t s y t l n a 2 55 u ² 2 i n c b 1 i

This preview shows page 6 - 11 out of 22 pages.

t ± 5 s   a t ² u t s   y t L N 0 , @ a 2 55 U ² @ / 2 I N " C B   " 1 I N " C B U   " 1   
Image of page 6

Subscribe to view the full document.

7 spatial trend: j s   ± longitude of site s k s   ± latitude of site s q t s   ± ) ² * j s   ² + k s   VI. Spatiotemporal models A. Introduction B. Modeling spatial correlation C. Modeling spatiotemporal correlation q t ± Nx 1   vector of unobserved factors for each state q t ± H 1 q t " 1 ² H 2 q t " 2 ² C ² H p q t " p ² 1 t Could have spatial correlation structure to 1 t WBC just take 1 t L N 0 , @ 1 2 I N   and set p ± 1
Image of page 7
8 q t ± Hq t " 1 ² 1 t We expect that: h sr ± 0 if r ± R s   and r p s WBC impose this outright. Here we’ll use a Bayesian prior that very strongly moves the data towards this without completely forcing: h sr | s p r , s ± R s   L N 0, A h 3 2   with A h 3 2 very small We also might expect h ss to be very similar (but not forced to be identical) for different s Let h 1 be a random variable that summarizes our prior subjective uncertainty about this common value: h 1 L N m h 1 , A h 1 2   where m h 1 is our prior guess about what h ss would be for a typical state and A h 1 2 , our uncertainty about this guess, might be large
Image of page 8

Subscribe to view the full document.

9 Finally, we may expect a pretty similar coefficient (represented by h 2 L N m h 2 , A h 2 2   ) relating q t s   to the average of neighboring states’ q t " 1 r   with A h 2 2 again large That is, the probability law for our prior on H can be represented as H ± h 1 I N ² h 2 B ² U h h 1 L N m h 1 , A h 1 2   h 2 L N m h 2 , A h 2 2   vec U h   L N 0 , A h 3 2 I N 2   Aside: if we wanted to say there is also a potential aggregate influence of the lagged value of all the other states combined (i.e., of 1 N U q t " 1 ), we could specify H ± h 1 I N ² h 2 B ² h 3 1 N 1 N U ² U h
Image of page 9
10 H ± h 1 I N ² h 2 B ² U h i N ± vec I N   b ± vec B   h ± vec H   ± h 1 i N ² h 2 b ² vec U h   prior: h | @ 1 2 L N m h , @ 1 2 M h   m h ± m h 1 i N ² m h 2 b M h ± A h 1 2 i N i N U ² A h 2 2 bb U ² A h 3 2 I N 2 Observed variable to be explained: y t ± N 1   vector of unemployment rates for each state Observed explanatory variables: w t ± k 1   vector of aggregate explanatory variables (common to all states) w t ± 1, t ,...   X t ± N d   matrix of explanatory
Image of page 10

Subscribe to view the full document.

Image of page 11
You've reached the end of this preview.
  • Winter '09
  • JamesHamilton
  • Unemployment, Trigraph, Yt, Qt, VI. Spatiotemporal models

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern