Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Addressing Critiques and Refining a regression Model

    Dear Profs and Colleagues,

    This model has faced criticism on several fronts (purple text), and I require assistance in understanding and addressing these critiques.
    Click image for larger version

Name:	2.png
Views:	1
Size:	3.4 KB
ID:	1733895

    The variable yi,s,d,t represents the outcome of a specific firm i in a particular sector s, district d, and year t. The time period under consideration spans from 2010 to 2019. The symbols Φt, Φs and Φd represent year, sector-specific and region-specific fixed effects, respectively.

    Firstly, in the context of human capital, it seems reasonable to consider human capital at the regional level as a control variable. If this is not taken into account, the proportion of immigrants at the regional level may inadvertently capture some of the broader effects of general human capital. What does it mean and how can I adjust the model according to this point?

    Secondly, there is a concern regarding the variation in dependent variables across firms, while explanatory variables vary within regions but remain constant for all firms within a given region. The mixed approach, involving regional factors in the explanatory matrix (X') and firm-specific factors in the dependent vector (y), is drawing criticism from referees due to the inclusion of fixed effects by region on the right-hand side (RHS) of the regressions. I seek guidance on how to interpret these critiques and make adjustments to the equation accordingly. I really don't know whats the problem. regional level? shall i remove the region from the equation? or ...

    Cheers,
    Paris

  • #2
    Use human capital as a regional control. Get human capital by region and add it to RHS. I don't even get the second criticism, so long as the REGIONAL predictors are time variant, your regional FE should be fine (in the sense that the model can, mathematically, estimate).

    Beyond this though, you give us no code, data, and even fewer context, so it's hard to say anything helpful. If this were my problem, I'd have my paper read by a coworker or your advisor.

    Comment

    Working...
    X