16 - Public Estimating Economic Impacts II Flashcards
(32 cards)
What was the empirical strategy in Approach #2?
A difference-in-differences (Diff-in-Diff) comparison of German cities that did and did not host 2006 World Cup matches.
What economic outcome was studied in Approach #2?
City-level unemployment.
What was the time range and data frequency for Approach #2?
Monthly data from Jan 1998 to Mar 2007, covering 75 cities.
What were the findings of Approach #2?
There was no statistically significant effect of hosting World Cup matches on unemployment.
Approach 2
What is the general equation for a Diff-in-Diff model?
The causal effect of the treatment (e.g., hosting a sports event).
Why must controls and covariates be included in Diff-in-Diff?
To account for other factors that could affect the outcome and avoid omitted variable bias.
What is the key assumption behind Diff-in-Diff?
Parallel trends — the treatment and control groups would have followed similar trends in the absence of treatment.
Why can’t you observe parallel trends post-treatment?
Because the treatment has already happened — so we must infer parallel trends based on pre-treatment data.
What is the implication of using repeated units over time?
OLS assumptions are violated unless we account for the panel nature of the data.
What was the empirical strategy in Approach #3?
Use ARIMA (Auto-Regressive Integrated Moving Average) models to test if events like the Pro Bowl affect Hawaii visitor numbers.
What data was used in Approach #3?
Daily airport arrivals (domestic + international) from Jan 1, 2004 to May 18, 2008.
What were the findings of Approach #3?
No statistically significant effect of events on visitor numbers.
What does an ARIMA model do?
Identifies regular patterns in a time series (like trends or seasonality) and tests whether events cause disturbances to those patterns.
Auto-Regressive Integrated Moving Average (ARIMA) models, first determine distinct time-series regularities (e.g. seasonality, underlying trends), and then see if events cause disturbances to the underlying regularities. A disturbance is a significant effect.
What type of data does ARIMA require?
Very frequent observations (e.g., daily data) to capture nuanced effects.
What is a potential reason for null results in sports economics?
Outcomes may be measured too infrequently or may suffer from measurement error.
By nature of hypothesis testing we are
generally concerned about Type I error.
Why can highly aggregated data be a problem?
It may mask local or short-term impacts, making it harder to detect real effects.
What does it mean to “overcontrol” in econometrics?
Including too many control variables (like fixed effects and trends) that may suppress real variation.
What happened when researchers studied minor NCAA championships without controls?
They found false significant effects, showing the importance of using appropriate controls.
What does this example teach us about overcontrolling?
Failing to control can produce misleading results, but proper controls often lead to accurate null findings.
If sports don’t have significant economic impact, should governments still subsidize them?
Possibly yes — if there are non-economic benefits like civic pride or health.
What are some examples of non-economic benefits from sports?
Community identity, Youth development, National/international prestige, Public health.
How can governments measure the value of non-economic benefits?
Using tools like: Contingent valuation surveys, Hedonic pricing models, Subjective well-being surveys.
