英语论文网

Capital structure theory predicts that firms with high growth opportunities should use less debt
So for a sample of firms run a regression of leverage (total debt over market value of firm) on Tobin’s q (a proxy for growth opportunities) where we observed q over several years for each firm

Panel Data

Residuals may be correlated across firms and/or across time
If so OLS standard errors are biased downwards
本论文由英语论文网 www.51lunwen.org整理提供

An Example
Capital structure theory predicts that firms with high growth opportunities should use less debt
So for a sample of firms run a regression of leverage (total debt over market value of firm) on Tobin’s q (a proxy for growth opportunities) where we observed q over several years for each firm

Big Problem I
A firm with a high q this year is likely to have a high q next year
The residuals of a given firm are correlated across year
Time series dependence
Firm effect
Solution is to use firm fixed effects or clustered (White) standard errors

Big Problem II
Firms within the same industry have similar levels of growth opportunties – q
The residuals of a given year are correlated across different firms
Cross-sectional dependence
Time effect
Solution is to use Fama-MacBeth
Runs T cross sectional regressions
The average of the T coefficient (variance) estimates is the coefficient (variance) estimate

Endogeneity
The independent variable is correlated with the error term.
Identification
Hausman test (p value)
Think of it as an intercept change
The regression coefficient may be biased up or down
Cause
Reverse causality between dependent and independent variable
Unobserved heterogeneity
Solution
Instrumental variables
Firm fixed effect – if unobserved heterogeneity is constant over time

An Example
本论文由英语论文网 www.51lunwen.org整理提供

Graham, Lemmon and Schallheim, 1998, JF
Capital structure theory predicts that firms with high tax rates should use more debt
So for a sample of firms run a regression of leverage (total debt over market value of firm) on Graham’s marginal tax rate estimate

Big Problem
If margin tax rate is high, theory predicts the firm will issue more debt, decreasing taxable income and thus decreasing the marginal tax rate
OLS - relation between debt and taxes is insignificant
The marginal tax rate is endogenous
Solution is to use instrumental variables
The chosen instrument is correlated with the explanatory variable but uncorrelated with the error term (dependent variable)
Estimated marginal tax rate for an all-equity firm
Observe a positive relation between debt levels and tax rates

Two-stage Least-squares (2SLS)
In the first stage, each endogenous variable is regressed on all valid instruments
Since the instruments are exogenous, these estimates of the endogenous variable will not be correlated with the error term
In the second stage, the regression is estimated as usual, except that in this stage each endogenous variable is replaced with its estimate in the first stage.

Sample Selection
The dependent variable is observed only for a restricted, nonrandom sample
The regression coefficient is biased
An intercept and 本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。