7,649
edits
Changes
Jump to navigation
Jump to search
===Instrumental Variable(s)===
See https://en.wikipedia.org/wiki/Instrumental_variables_estimation for background.
We want a variable that affects the amount of venture capital only through its effect on agglomeration. More specifically, in the case of the highest1hull, we want a variable that affects the next periods VC only through its effect on the hull size (given that there is always one and only one hull). Moreover, we will be using city-level fixed effects, so a variable that doesn't change over time -- like the size of the city -- will have no variation to drive an effect on hull size. There are two immediate possibilities, using a size measure involving the TIFs or using the number of locations within TIFs. The points measure has two opposing effects in it: it grows as more startups move into TIFs and it grows as TIFs take up more area. The size measure(s) could be the area of TIfs intersecting last periods hull and the area not intersecting it.
hull area in t = f( hull area in t-1 intersecting TIF areas in t, TIF areas in t not intersecting hull area in t-1).
The two measures will have temporal variation as TIF areas vary over time and as hull areas vary over time...
Urban Start-up Agglomeration and Venture Capital Investment (view source)
Revision as of 13:19, 23 October 2019
, 13:19, 23 October 2019no edit summary
=New Work=
==Target Journal==
Options might include:
*Science: See https://science.sciencemag.org/content/347/6222/606
*Management Science
==Files==
Still to do:
*ACS variables- Just needs a join!
*Run and add in chosen layers!
==Analysis Exploration==
Choose scale regressors:
*growthinv17l numdealsl numstartupsl
Then the analysis will be loosely as follows:
*Non-layer dependent descriptives
*Explain layers using Burlington, VT?
*Highest1Hull
*Elbow
*Layer Dependent descriptives I
*Maximum R2
*Layer Dependent descriptives II
*Group Means
*Policy Simulation
===Base specification for highest1hull===
As a consequence we can't use nohull, as it is always 1. It also turns out that tothullcount and tothulldensity don't matter, likely because we already include numstartupsl (removing it makes tothullcount significant). avgdisthm, however, does work.
===Instrumental Variable(s)===
See https://en.wikipedia.org/wiki/Instrumental_variables_estimation for background.
We want a variable that affects the amount of venture capital only through its effect on agglomeration. More specifically, in the case of the highest1hull, we want a variable that affects the next periods VC only through its effect on the hull size (given that there is always one and only one hull). Moreover, we will be using city-level fixed effects, so a variable that doesn't change over time -- like the size of the city -- will have no variation to drive an effect on hull size. There are two immediate possibilities, using a size measure involving the TIFs or using the number of locations within TIFs. The points measure has two opposing effects in it: it grows as more startups move into TIFs and it grows as TIFs take up more area. The size measure(s) could be the area of TIfs intersecting last periods hull and the area not intersecting it.
hull area in t = f( hull area in t-1 intersecting TIF areas in t, TIF areas in t not intersecting hull area in t-1).
The two measures will have temporal variation as TIF areas vary over time and as hull areas vary over time...
===Instrumenting highest1hull===
*Endogeneity test is 0.037 with Chi-sq(1) P-val = 0.8465. Null is that it is exogeneous, which we can't reject.
Overall, though, it looks like the TIF instrument is dead. Which is a HUGE shame.
==CoLevelForCircles==
