Difference between revisions of "Hornbeck (2010)"
imported>Moshe |
|||
| (16 intermediate revisions by 3 users not shown) | |||
| Line 1: | Line 1: | ||
| + | {{Article | ||
| + | |Has page=Hornbeck (2010) | ||
| + | |Has bibtex key= | ||
| + | |Has article title= | ||
| + | |Has author=Hornbeck | ||
| + | |Has year=2010 | ||
| + | |In journal= | ||
| + | |In volume= | ||
| + | |In number= | ||
| + | |Has pages= | ||
| + | |Has publisher= | ||
| + | }} | ||
Return to [[BPP Field Exam Papers 2012]] | Return to [[BPP Field Exam Papers 2012]] | ||
==Empirical Questions: == | ==Empirical Questions: == | ||
| Line 8: | Line 20: | ||
The effect on Investment from a change in cost of protection equals the change in Investment from a change in protection multiplied by the change in protection from a change in cost of protection. | The effect on Investment from a change in cost of protection equals the change in Investment from a change in protection multiplied by the change in protection from a change in cost of protection. | ||
| − | Since <math>\frac{\partial P}{\partial C_{p}} < 0</math> we know that an estimate of <math>\frac{\partial I}{\partial C_{p}}</math> is informative about the sign of <math>\frac{\partial P}{\partial C_{p}}</math> | + | Since <math>\frac{\partial P}{\partial C_{p}} < 0</math> we know that an estimate of <math>\frac{\partial I}{\partial C_{p}}</math> is informative about the sign of <math>\frac{\partial I}{\partial P}</math> |
| + | |||
| + | So, we can think of <math>\frac{\partial I}{\partial C_{p}}</math> as the "reduced form" where marginal cost of protection is an instrumental variable. Since we do not have data on protection levels, we can not estimate the "first stage" and recover <math>\frac{\partial I}{\partial P}</math>. | ||
=== How does the author test the hypothesis? === | === How does the author test the hypothesis? === | ||
| + | The author's take a difference in difference approach where the main specification is: | ||
| + | |||
| + | <math>Y_{ct}-Y_{c(t-1)}= \alpha_{st} + \beta_{1t}W_{c}+\beta_{2t}W_{c}^{2}+\beta_{3t}W_{c}^{3}+\beta_{4t}W_{c}^{4}+ \epsilon_{ct}</math> | ||
| + | *the estimated <math>\beta</math> are allowed to vary in each decade and summarize how changes over each decade in county outcome Y vary by country woodland level W | ||
| + | |||
| + | * We are looking at low woodland areas vs high woodland areas. The sample is county & decade. We need the introduction of barbwire to be endogenous i.e. it can't be correlated with better crops in any way. | ||
| + | * We use change in costs as an IV for change in protection. | ||
=== What do the authors tests achieve?=== | === What do the authors tests achieve?=== | ||
| + | Findings include that from 1880 to 1890 and 1890 to 1900 counties with the least woodland made large relative gains in the improvement of farmland. By contrast, there were no substantial changes at low woodland areas before 1880 or after 1900 or in higher woodland areas from 1880-1900. | ||
===How could the tests be improved on? Strengths? Weaknesses?=== | ===How could the tests be improved on? Strengths? Weaknesses?=== | ||
| Line 20: | Line 42: | ||
=== How does the author rule out alternative hypotheses? === | === How does the author rule out alternative hypotheses? === | ||
| + | *Counties with less woodland tend to be further west, so there is concern that estimates could be measuring some push towards increased western development. Results are robust to control for distance west of StL. | ||
| + | *There is also a worry that counties with different woodland levels may be suited to different crops, so changes in prices/technologies contribute to differential growth. Merge data with soil data and allow for differential quadratic growth trends. | ||
| + | *Convergence? The author adds controls for 1870 level of outcome interacted with each decade to allow for differential growth. | ||
| + | *Construction of Railroad lines my be endogenous and effect barbwire development. Add controls in country railroad track mileage. | ||
Latest revision as of 18:14, 29 September 2020
| Article | |
|---|---|
| Has bibtex key | |
| Has article title | |
| Has author | Hornbeck |
| Has year | 2010 |
| In journal | |
| In volume | |
| In number | |
| Has pages | |
| Has publisher | |
| © edegan.com, 2016 | |
Return to BPP Field Exam Papers 2012
Contents
- 1 Empirical Questions:
- 1.1 What is the author's topic and hypothesis?
- 1.2 How does the author test the hypothesis?
- 1.3 What do the authors tests achieve?
- 1.4 How could the tests be improved on? Strengths? Weaknesses?
- 1.5 What are some alternative empirical strategies
- 1.6 How does the author rule out alternative hypotheses?
Empirical Questions:
What is the author's topic and hypothesis?
This paper examines the impact on agricultural development from a decrease in the cost of protecting farmland. Barbed wire appears to have had a substantial impact on agriculture development in the US and in particular, this may reflect an important role for protecting land and securing farmers' full bundle of property rights.
Theoretical Framework: [math]\frac{\partial I}{\partial C_{p}}=\frac{\partial I}{\partial P} \cdot \frac{\partial P}{\partial C_{p}}[/math]
The effect on Investment from a change in cost of protection equals the change in Investment from a change in protection multiplied by the change in protection from a change in cost of protection.
Since [math]\frac{\partial P}{\partial C_{p}} \lt 0[/math] we know that an estimate of [math]\frac{\partial I}{\partial C_{p}}[/math] is informative about the sign of [math]\frac{\partial I}{\partial P}[/math]
So, we can think of [math]\frac{\partial I}{\partial C_{p}}[/math] as the "reduced form" where marginal cost of protection is an instrumental variable. Since we do not have data on protection levels, we can not estimate the "first stage" and recover [math]\frac{\partial I}{\partial P}[/math].
How does the author test the hypothesis?
The author's take a difference in difference approach where the main specification is:
[math]Y_{ct}-Y_{c(t-1)}= \alpha_{st} + \beta_{1t}W_{c}+\beta_{2t}W_{c}^{2}+\beta_{3t}W_{c}^{3}+\beta_{4t}W_{c}^{4}+ \epsilon_{ct}[/math]
- the estimated [math]\beta[/math] are allowed to vary in each decade and summarize how changes over each decade in county outcome Y vary by country woodland level W
- We are looking at low woodland areas vs high woodland areas. The sample is county & decade. We need the introduction of barbwire to be endogenous i.e. it can't be correlated with better crops in any way.
- We use change in costs as an IV for change in protection.
What do the authors tests achieve?
Findings include that from 1880 to 1890 and 1890 to 1900 counties with the least woodland made large relative gains in the improvement of farmland. By contrast, there were no substantial changes at low woodland areas before 1880 or after 1900 or in higher woodland areas from 1880-1900.
How could the tests be improved on? Strengths? Weaknesses?
What are some alternative empirical strategies
How does the author rule out alternative hypotheses?
- Counties with less woodland tend to be further west, so there is concern that estimates could be measuring some push towards increased western development. Results are robust to control for distance west of StL.
- There is also a worry that counties with different woodland levels may be suited to different crops, so changes in prices/technologies contribute to differential growth. Merge data with soil data and allow for differential quadratic growth trends.
- Convergence? The author adds controls for 1870 level of outcome interacted with each decade to allow for differential growth.
- Construction of Railroad lines my be endogenous and effect barbwire development. Add controls in country railroad track mileage.
