S11 - Time Series & Panel Data Flashcards
What are the Basics of a Time-Series?
Time: Clear temporal ordering and is indexed by time
Sampling: Not a random sample but it is still the outcome of stochastic process
Population: The set of all possible realizations of the time series
What is the difference from cross-sections to time-series?
In a standard cross-sectional set up (normal wie gehabt?), we implicitly assume that an effect of x on y is instantaneou
> Time-series allows for delayed effects or effects that persist over time
-> Cross-sections & Time-series together = panel data
Which assumptions do time series violate? What is the problem?
Time-series commonly violate:
> mean independence (2) and
> the independence of errors (4)
Problem: we need large samples, so that the model will still produce unbiased estimators
-> very difficult to get a long time-series (most difficult to collect this data)
The Autocorrelation plot is primarily used to diagnose whether a Time Series violates the crucial non-stationarity assumption.
Select one:
True
False
This is FALSE. The plot is primarily used to diagnose whether a Time Series is weakly dependent, whether a Time Series violates the crucial persistence assumption.
The correct answer is ‘False’.
A problem with time series is that many variables are positively or negatively trended, which increases the risks of spurious regression results.
Select one:
True
False
This is TRUE. If two variables are time trended, there will be a relationship between them. But this relationship is spurious, if it is merely driven by the common time trend.
The correct answer is ‘True’.
A researcher seeks to investigate the relationship between temporary membership of the UN Security Council (every year five countries are elected to the Security Council for a two-year term) and development aid from the United States. She runs a regression where the independent variable of interest is a dummy variable for whether a country was a member of the UN Security Council in a given year, and the dependent variable is the amount of aid received from the United States in the same year. The analysis uses 50 years of annual data from 161 developing countries across the world.
What kind of data is this?
a. Cross-sectional data
b. Time series data
c. Panel data
Your answer is incorrect.
Time series data: single unit over multiple time periods
Cross-sectional data: multiple units over single time period
Panel data: multiple units over multiple time periods.
The correct answer is: Panel data
Indicate whether the following statement is true or false:
An OLS analysis of panel data using dummy variables representing each country in the dataset will yield the same coefficients and standard errors on the independent variables as a fixed effects model run on the same dataset.
Select one:
True
False
True
Select all answers below that are TRUE
a.
Fixed effects allow a different y-intercept for each panel unit
b.
Fixed effects models are also known as “within” models because they only use variation within the panel units but not between panel units
c.
Fixed effects models are subject to omitted variable bias from leaving out control variables that vary over time
d.
Coefficients from fixed effects models are more frequently biased than those from pooled cross-sectional models.
e.
Fixed effects models estimate the same regression coefficient for a given variable in all panel units (e.g., countries).
The correct answers are: Fixed effects models are subject to omitted variable bias from leaving out control variables that vary over time, Fixed effects models are also known as “within” models because they only use variation within the panel units but not between panel units, Fixed effects models estimate the same regression coefficient for a given variable in all panel units (e.g., countries)., Fixed effects allow a different y-intercept for each panel unit
Non-Stationarity can have two reasons, namely…
Time Trends
Unit Roots
What is the problem with time trends?
Spurious regression
What can be seen with a random walk? Whats the cure?
The variance of a random walk increases as a linear fn of time, therefore it cannot be stationary
differencing
What does stationarity mean?
A time-series is (weakly) stationary if its mean and variance remain constant over time.
de facto, this implies that you should see
I mean reversion
I frequent crossing of the mean line
I no trend
What are the two crucial presumptions that must be given in time-series?
stationarity
weakly dependence
Definition of weakly dependent Time Series
A covariance stationary time-series is weakly dependent if the correlation between xt and xt+h goes to zero “sufficiently quickly” as h increases.
WHat can serial correlation lead to?
Serial correlation in the errors can make them appear smaller than they should be (type I error) but the coe cient estimates are usually still unbiased or at least consistent.
