Katrina and the Gulf States Casino Industry

1
Katrina and the Gulf States Casino Industry

• Douglas M. Walker (College of Charleston)
• and John D. Jackson (Auburn University)
• Paper presented at the 2007 Southern Economic
  Association Conference, November 19-21, 2007, New
  Orleans, LA.

2
Background

• Katrina hit along the LA-MS border in Aug. 2005.
• Rita hit near the LA-TX border in Sept. 2005.
• The casino industry in LA and MS was devastated,
  since much of the industry was located in
  hurricane affected areas.
• Our analysis is of the relationship between the
  casino industry and state-level personal income,
  and how this relationship was affected by
  Hurricanes Katrina and Rita.

3
Figure 1. Maps of casino locations in Louisiana
and Mississippi
Figure 1. Maps of casino locations in Louisiana
and Mississippi
Source American Gaming Association (2006b)
4
The casino industry in LA and MS

• LA and MS are relatively isolated markets for
  commercial casinos
• They attract tourism
• Isolation
• Situated on state borders/rivers, or in tourist
  cities
• Casinos may drive state-level economic growth
• Walker and Jackson (1998) found a significant
  short-run effect using 1991-96 data, for all U.S.
  states with commercial casinos
• But our 2007 paper found no significant long-term
  effect, using 1991-2005 data
• Econometric studies of the effects of casino
  gambling are still rare

5
Table 1. Casinos damaged by Katrina and Rita
6
Figure 2. Nominal casino revenue, N.O. area
casinos affected by Katrina (millions of )

• Source Louisiana Gaming Control Board

7
Figure 3. Nominal casino revenue, Lake Charles
area casinos affected by Rita (millions of )

• Source Louisiana Gaming Control Board

8
Table 2. Mississippi Gulf Coast casino slot
machine positions, casinos affected by Katrina
9
Our question

• The previous tables and charts indicate the how
  quickly the casino industry rebuilt.
• Would such a strong recovery in this industry
  have an impact on the states post-Katrina (and
  Rita) economies?

10
The data

• We use real quarterly LA and MS personal income
  and casino revenue data.
• Quarterly data are used rather than annual, in
  order to maximize the number of observations.
• Post-Katrina annual data are limited
• Our casino revenue data exclude non-casino video
  poker and slot machines.
• Indian casinos are excluded because data are not
  publicly available.

11
Data, cont.

• We use personal income rather than per capita
  income to measure economic growth.
• Before and after Katrina there was significant
  migration from New Orleans and the
  Biloxi/Gulfport area.
• Migration may not be accurately reflected in per
  capita income data.
• Personal income data are available quarterly
  state-level per capita GDP data are not.

12
Data, cont.

• Data set begins in 1997.1
• We want to avoid picking up the initial stimulus
  effect we found in our 1998 paper.
• The Louisiana Gaming Control Board was created in
  1996 previously the State Police reported data.
  Beginning in 1997 allows us to use data from only
  one agency.
• Aside from these reasons, beginning in 1997 is a
  nice balance to the post-Katrina data available.

13
Data, cont.

• The data end in 2006.3, the last period for which
  personal income data were available (as of the
  paper writing).
• 39 quarters of data on each variable for each
  state, for 78 total observations.
• Katrina and Rita were both in 2005.3.
• 34 observations in each state are pre-hurricane
  (1997.1-2005.2)
• 5 post-hurricane observations (2005.3-2006.3)

14
The model

• Personal income (t0) b1 Constant b2 Personal
  income (t-4) b3 Casino revenue (t0) b4 Casino
  revenue (t-4) b5 Katrina dummy b6 Katrina
  (Casino revenue) b7 State dummy e (t0)
• Personal income one year prior is an explanatory
  variable since annual personal income is a random
  walk.
• A four period lag for casino revenues was chosen
  after an analysis of the correlogram.
• Current values for casino revenue appear not to
  be affected by immediate past quarters.
• The fourth lagged period of casino revenue
  significantly affects current periods.
• Probably an annual cycle.
• State dummy is included (1 for MS, 0 for LA)
• Katrina dummy (1 from periods 2005.3 through
  2006.3 0 otherwise) includes effect from Rita

15
Model, cont.

• Katrina revenue interaction term will pick up
  relationship between casino gambling and personal
  income above and beyond that explained by revenue
  t0 and t-4
• This should tell us how the effect of casino
  revenues on income enhances (b6gt0) or detracts
  (b6lt0) from economic recovery.
• Omitted variables include
• Seasonal dummies insignificant, and didnt
  affect other coefficients
• Time trend insignificant and didnt affect other
  coefficients
• FEMA relief, other federal or state relief, and
  insurance settlements were not explicitly
  accounted for in the model
• BEA indicates that these are accounted for in the
  2005.3 personal income data

16
Table 3. Regression results. Dependent variable
Personal income (adj. for inflation)
17
Conclusions

• Clearly, the hurricanes devastated personal
  income in LA and MS
• The hurricane dummy is strongly negative
• The interaction term is positive, indicating that
  increases in casino revenues have a significantly
  greater expansionary effect on personal income
  after the hurricanes than before them.
• This result is consistent with our earlier
  findings (1998) of at least a short-term growth
  effect from casino gambling.
• Capital investment, labor flows, etc., may
  explain this.
• Copies of the paper available at
  http//www.cofc.edu/walkerd