Scribd
Upload
39 views19 pages

Endogeneity

The document discusses challenges to causal inference in corporate research and how shock-based causal inference approaches can help address these challenges. It outlines 5 conditions a "good shock" should meet for credible causal inference: 1) the shock is strong enough to change behavior, 2) the shock is exogenous, 3) assignment to treatment is "as if random", 4) there is covariate balance between treated and control groups, and 5) the only effect is through the specific channel of the shock. Examples of shocks that can be used include natural disasters, regulatory changes, and market competition changes.

Uploaded by

Rabeya Aktar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
39 views19 pages

Endogeneity

The document discusses challenges to causal inference in corporate research and how shock-based causal inference approaches can help address these challenges. It outlines 5 conditions a "good shock" should meet for credible causal inference: 1) the shock is strong enough to change behavior, 2) the shock is exogenous, 3) assignment to treatment is "as if random", 4) there is covariate balance between treated and control groups, and 5) the only effect is through the specific channel of the shock. Examples of shocks that can be used include natural disasters, regulatory changes, and market competition changes.

Uploaded by

Rabeya Aktar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Endogeneity

Overview
Causation

What is endogeneity

Reasons for endogeneity

Shock-based causal inference

Conditions of good shock

Different Approaches for shock-based designs


Shock-Based Causal Inference in Corporate

Finance and Accounting Research

Vladimir Atanasov

Bernard Black

Critical Finance Review, 2016, 5: 207–304


• Corporate finance research is concerned with causation:

- Does a change in some input cause a change in some output?


- Does corporate governance affect firm performance?
- Does capital structure affect firm investments?
- How do corporate acquisitions affect the value of the acquirer, or the acquirer and target
together?

• Without a causal link, we lack a strong basis for recommending that


firms/stakeholders/Govt. change their behavior.

• For example, corporate governance research. Decisionmakers – corporate boards,


investors, regulators – want to know whether a change in governance will cause a change
in firm value or performance.
To provide a credible basis for “causal inference” (sometimes called “identification”), a research
design must address multiple econometric concerns.

Some of these are referred to as “endogeneity”.


Empirical Challenges to Causal Inference in Corporate Governance Research

Suppose we run such a regression, using ordinary least squares (OLS):

qi = α + β1 ∗ govi + β2 ∗ xi +εi

From this equation, we cannot infer that a change in gov will cause a change in q – using
“cause” to mean that, if one increases gov, changing nothing else, q will increase (if β1 is
positive and significant).
Potential Problems are:

1. Reverse causation/causality

Reverse causality means that X and Y are associated, but not in the way we would expect. While
X causing a change in Y, it can be the other way also: Y is causing changes in X.

For our corporate governance example, perhaps q causes gov. So, regression cannot tell us the
direction of the causal arrow.
“…one may be tempted to say that low social status causes schizophrenia, [but] another
plausible explanation is that schizophrenia causes downward social mobility…” - Gerstman
2. Omitted variable bias

• Omitted-variable bias (OVB) occurs when a statistical model leaves out one or more
relevant variables.

• For our corporate governance example, perhaps one or more unobserved variables
cause both q and gov, or mediate the relationship between q and gov. Without them,
the coefficient β1 is a biased estimate of the true causal effect of gov on q.
3. Measurement error
i. Specification Error

Even if we could perfectly measure X and all relevant covariates, we would not know the
functional form through which each influences Y. The missing part of the correct
specification leads to biased coefficients, just like any other omitted variable.

ii. Construct validity

The construction of variables may poorly fit the underlying concept.


4. Simultaneity

Bidirectional causation, with X causing Y, and Y also causing X. OLS regression


will provide a biased estimate of the magnitude and perhaps the sign of the effect.

5. Heterogeneous Effects

6. Observation bias

Observed subjects behave differently because they are observed. Firms which
change gov may behave differently because their managers or employees think the
change in gov matters, when in fact it has no direct effect.
7. Interdependent effects

Interdependent effects on firms which adopt a reform. For example, a governance reform that
will not affect share price for a single firm might be effective if adopted widely, because
investors will then appreciate the reform’s impact. Conversely, a reform which improves
efficiency for a single firm might not improve profitability if adopted widely, because the gains
will be competed away.

These obstacles to credible causal inference


Shock-Based Causal Inference

Shock-based designs use an external shock to limit endogeneity. At their best, they can
approach, but never achieve, fully random assignment. Different designs are used for this
purpose:
i. Difference-in-Difference (DiD)
ii. Event Study (ES)
iii. Instrumental Variable (IV)
iv. Regression Discontinuity (RD)
DiD, ES, IV, and RD – appear to rely on different assumptions. However, we will argue,
these designs share core elements. All rely on a “good shock” – one which permits
credible causal inference.
A good shock should satisfy five conditions:

(1) Shock Strength: The shock is strong enough to significantly change firm behavior or
incentives.

(2) Exogeneous Shock: The shock came from “outside” the system one is studying.
Treated firms did not choose whether to be treated, cannot anticipate the shock. If the
shock is exogenous, we are less worried that unobservable might be correlated with
both assignment to treatment and the potential outcomes, and thus generate omitted
variable bias. Shock exogeneity should be defended, not just assumed.

(3) “As If Random” Assignment: The shock must separate firms into treated and controls
in a manner which is close to random. Different research designs can accommodate
different depar- tures from random assignment, but the closer the shock comes to
random assignment, the more credible it will be.
4. Covariate balance. The forcing and forced variables aside, the shock
should produce reasonable covariate balance between treated and
control firms, including “common support” (reasonable overlap
between treated and control firms on all covariates). Somewhat
imperfect balance can be address with balancing methods, but severe
imbalance undermines shock credibility, even if the reason for
imbalance is not obvious. Covariate balance should be reported.
5. Only-Through Condition(s): We must have reason to believe that the
apparent effect of the shock on the outcome came only through the
shock (sometimes, through a specific channel). The shock must be
“isolated” – there must be no other shock, at around the same time,
that could also affect treated firms differently than control firms. And if
one expects the shock to affect outcomes through a particular channel,
the shock must also affect the outcome only through that channel. In
IV analysis, this is called an “exclusion restriction,” because one
assumes away (excludes) other channels; we prefer the more
descriptive term “only-through condition.”

Conditions (1) and (2) are part of standard discussions of DiD, and (1), (2) and (5) are well-known for IV
The Deepwater Horizon Oil Spill

2008–2009 financial crisis

Peer Market Competiion

Political/ Policy uncertainty

Legal change

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy