BƯỚC 3. ROBUSTNESS, ALTERNATIVE TUI VÀ ALTERNATIVE DIVIDEND MEASURES

PHẦN A. ROBUSTNESS PLAN

Concern	Test
TUI aggregation	annual vs fiscal-year-aligned
TUI scale	log / standardized
generic uncertainty	EPU/GPR/WUI controls
dividend metric	DVC/AT, payer, payout, yield
sticky dividends	change/cut/omission
repurchase substitution	total payout/repurchase
tourism classification	core vs broad
country dominance	leave-one-country-out
crisis dependence	COVID exclusion
inference	country cluster / wild bootstrap
sector cycles	Industry×Year FE

PHẦN B. FISCAL-YEAR TUI

One of the most important tests.

Compare:

$$
AnnualTUI_{c,t}
$$

vs:

$$
TUIFY_{i,c,t}
$$

If fiscal-year TUI is stronger:

timing alignment matters.

Report FYE distribution.

PHẦN C. TUI TRANSFORMATIONS

Possible:

• raw TUI;
• standardized TUI;
• $ln(1+TUI)$ where valid;
• high-TUI dummy;
• ΔTUI.

Do not transform only to get significance.

PHẦN D. GENERAL UNCERTAINTY CONTROLS

Add sequentially:

• EPU;
• GPR;
• WUI;
• VIX/global uncertainty.

If TUI remains:

supports tourism-specific information.

If TUI vanishes:

TUI may proxy broader uncertainty.

This is substantively important.

PHẦN E. ALTERNATIVE DIVIDEND OUTCOMES

1. DVC/AT.
2. payer dummy.
3. payout ratio.
4. DVC/SALE.
5. dividend yield.
6. dividend growth.
7. cut.
8. omission.
9. total payout if repurchase data validated.

PHẦN F. COVID SENSITIVITY

Tourism uncertainty and dividends both changed dramatically in 2020–2021.

Tests:

• exclude 2020;
• exclude 2020–2021;
• pre-COVID;
• post-COVID;
• TUI×COVID.

Do not let one extraordinary episode drive whole story.

PHẦN G. COUNTRY LEAVE-ONE-OUT

Essential if coverage is concentrated.

Excluded country	β	p	N	sign

Large countries/tourism markets may dominate.

PHẦN H. TOURISM SUBSECTOR

Compare:

• airlines;
• hotels;
• restaurants;
• travel;
• recreation/casino.

Expected uncertainty sensitivity can differ sharply.

PHẦN I. VCE

Because TUI is country-year:

• country cluster baseline;
• wild-cluster if few countries;
• country/year two-way sensitivity;
• firm+country sensitivity where justified.

Do not choose firm clustering as sole inference.

PHẦN J. ROBUSTNESS SUMMARY

Family	Sign	Magnitude	Precision	Judgment
TUI timing
TUI transformations
uncertainty controls
dividend outcomes
COVID
tourism classification
LOO
VCE

PHẦN K. SAMPLE WRITING

The main result is broadly robust to alternative TUI timing and payout definitions. Fiscal-year-aligned TUI yields [similar/stronger/weaker] estimates than calendar-year TUI. The coefficient remains [direction] after controlling for broader EPU, GPR, and global uncertainty, suggesting that tourism-specific uncertainty contains incremental information. The conclusion is also [stable/sensitive] to excluding the COVID-19 period and individual countries.

PHẦN L. OUTPUT

1. Annual vs Fiscal-Year TUI Table.
2. General-Uncertainty Controls Table.
3. Alternative Dividend Table.
4. COVID Table.
5. Subsector Table.
6. Country LOO.
7. VCE Table.
8. Robustness Summary.
9. Draft robustness 1,500–2,200 words.

PHẦN M. CHECKLIST

☐ fiscal-year TUI tested.
☐ generic uncertainty controlled.
☐ multiple payout measures used.
☐ COVID sensitivity done.
☐ subsectors examined.
☐ country LOO done.
☐ VCE robust.
☐ no significance shopping.

Chuyển sang Bước 4: Endogeneity & Advanced Analyses.