BƯỚC 1. XÂY DỰNG PHƯƠNG PHÁP NGHIÊN CỨU

Bài mẫu: U.S.–China Tension and Corporate Capital Structure in Southeast Asia
Mục tiêu đầu ra: Hoàn thành bản nháp có căn cứ tài liệu tham khảo cho phần Data and Methodology của paper.

Quy ước của bài mẫu

• Biến độc lập chính: U.S.–China Tension Index (UCT) của Rogers, Sun, and Sun (2024).
• UCT là một news-based global time-series index, không phải country-specific index.
• Dữ liệu UCT được lấy từ Economic Policy Uncertainty / PolicyUncertainty.com.
• Dữ liệu tài chính doanh nghiệp: Compustat North America + Compustat Global, nhưng sample nghiên cứu chỉ giữ các firms thuộc Đông Nam Á.
• Đơn vị quan sát: firm-year.
• Universe địa lý có thể bao phủ các nền kinh tế Đông Nam Á/ASEAN; final sample phụ thuộc data availability.
• Core empirical sample khuyến nghị nếu dùng Compustat: Indonesia, Malaysia, Philippines, Singapore, Thailand, and Vietnam.
• Financial firms và utilities bị loại.
• Yêu cầu:

$$
AT>0
$$

và:

$$
SALEgeq0
$$

• Continuous dependent variables và firm-level controls được winsorize ở 1st–99th percentiles; macro variables và UCT không winsorize trong baseline.
• Main Capital Structure measure: Book Leverage.
• Alternative Capital Structure measures:
  1. Short-Term Leverage;
  2. Long-Term Leverage;
  3. Market Leverage;
  4. Debt Maturity.
• Main theoretical prediction: higher U.S.–China tension may reduce firms’ willingness or ability to use debt, especially short-term debt, but the sign is ultimately an empirical question because uncertainty may also increase external financing needs for some firms.
• Critical identification issue: annual UCT varies only over time. Therefore, if annual UCT is identical for all firms in year $t$, Year Fixed Effects absorb the main effect of UCT completely.

• The guide therefore distinguishes three valid empirical designs:

  Design A. Annual UCT direct-effect model without Year FE
  Design B. Fiscal-year-aligned UCT using monthly UCT, potentially allowing Year FE if sufficient within-year variation remains
  Design C. Exposure-based model: UCT × U.S./China Exposure with Year FE

Không được chạy:

$$
CapitalStructure_{i,t}

beta UCTt
+
FirmFE
+
YearFE
+
varepsilon{i,t}
$$

nếu $UCT_t$ chỉ có một giá trị chung cho toàn bộ firms trong từng năm, vì $UCT_t$ sẽ bị Year FE hấp thụ.

---

PHẦN A. MỤC TIÊU CỦA BƯỚC

1. Sau Bước 1 phải trả lời được những câu hỏi nào?

Sinh viên phải xác định đầy đủ:

1. U.S.–China Tension Index lấy từ đâu?
2. UCT được xây dựng như thế nào?
3. UCT có variation ở cấp nào?
4. UCT là monthly hay annual trong dataset được cung cấp?
5. UCT được ghép với fiscal-year firm data như thế nào?
6. Vì sao UCT không phải country-year variable?
7. Vì sao annual UCT bị absorbed bởi Year Fixed Effects?
8. Có thể xử lý vấn đề Year FE bằng những design nào?
9. Dữ liệu firms Đông Nam Á lấy từ đâu?
10. Nghiên cứu dùng toàn bộ Đông Nam Á hay ASEAN-6?
11. Những countries nào thực sự có trong final sample?
12. Tại sao loại financial firms?
13. Tại sao loại utilities?
14. Tại sao yêu cầu AT > 0?
15. Tại sao yêu cầu SALE ≥ 0?
16. Missing observations được xử lý thế nào?
17. Biến nào được winsorize?
18. UCT có winsorize không?
19. Capital Structure được đo bằng những variables nào?
20. Main Book Leverage formula là gì?
21. Short-Term Leverage formula là gì?
22. Long-Term Leverage formula là gì?
23. Debt Maturity được đo như thế nào?
24. Market Leverage được construct như thế nào?
25. Tại sao cần tách short-term và long-term debt?
26. Controls được lựa chọn từ capital-structure literature nào?
27. Firm controls và macro controls gồm những gì?
28. Trade-off theory dự đoán gì?
29. Pecking-order theory dự đoán gì?
30. Uncertainty có thể ảnh hưởng debt financing qua những channels nào?
31. Có paper nào trực tiếp nghiên cứu UCT → leverage chưa?
32. Paper trực tiếp về Việt Nam năm 2026 tìm thấy gì?
33. Vì sao Đông Nam Á là setting phù hợp?
34. U.S.–China trade fragmentation có liên quan gì tới ASEAN?
35. Baseline equation là gì?
36. FE nào được phép dùng trong từng design?
37. VCE phải xử lý thế nào khi UCT chỉ thay đổi theo time?
38. Có cần cluster theo year không?
39. Có cần two-way clustering không?
40. Có few-year-cluster concern không?
41. Exposure-based design có thể dùng exposure nào?
42. Country-level trade exposure khác firm-level exposure ở điểm gì?
43. Heterogeneity nào nên chuẩn bị từ Bước 1?
44. Mọi methodological decision đã có citation chưa?

Thành phẩm cuối Bước 1 phải đủ để viết Section 3. Data and Methodology của paper.

---

PHẦN B. DATA SOURCES

2. Dữ liệu tài chính doanh nghiệp

Bài mẫu sử dụng dữ liệu accounting và market data từ:

• Compustat Global;
• Compustat North America chỉ được giữ trong master data nếu workflow chung sử dụng cả hai, nhưng final Southeast Asia sample chủ yếu đến từ Compustat Global.

Đơn vị quan sát:

firm-year

Sinh viên phải xác định

Nội dung	Kết quả cần điền
Database
Sample period
Unit of observation
Firm identifier
Fiscal-year variable
Fiscal-year-end date
Country identifier
Industry identifier
Industry classification
Number of firms
Number of firm-year observations
Number of countries

Không được tự suy đoán

• firms;
• observations;
• countries;
• years.

Phải lấy từ:

• data summary;
• Stata output;
• sample-selection output;
• final regression sample.

---

PHẦN C. U.S.–CHINA TENSION INDEX

3. Nguồn UCT

Main independent variable:

U.S.–China Tension Index (UCT)

Nguồn:

Rogers, J. H., Sun, B., & Sun, T. (2024). U.S.–China Tension.

Working paper / SSRN:

DOI: 10.2139/ssrn.4815838

Data source:

Economic Policy Uncertainty / PolicyUncertainty.com

Citation được khuyến nghị khi dùng series

Rogers, J. H., Sun, B., & Sun, T. (2024), “U.S.-China Tension.”

---

4. UCT được xây dựng như thế nào?

Rogers et al. xây dựng UCT bằng newspaper-based textual analysis.

Index theo dõi tỷ trọng các bài viết có đồng thời:

1. mention of the United States/U.S.;
2. mention of China/Chinese;
3. contentious issues trong bilateral relationship;
4. tension-related phrases.

Search terms được hỗ trợ bởi topic-modeling approaches như:

• K-means;
• guided LDA;
• Newsmap.

Methodology có nền tảng từ newspaper-based uncertainty measurement của Baker, Bloom, and Davis (2016).

Interpretation

$$
Higher UCT

Greater U.S.text{–}China Tension
$$

Economic meaning

UCT phản ánh tensions liên quan tới:

• trade disputes;
• tariffs;
• sanctions;
• technology restrictions;
• investment restrictions;
• geopolitical disputes;
• diplomatic confrontation;
• supply-chain concerns.

Đoạn viết mẫu

We measure U.S.–China tension using the newspaper-based index developed by Rogers et al. (2024). The index tracks the share of articles in leading U.S. newspapers that jointly mention the United States and China, contentious bilateral issues, and phrases indicating heightened tension. Higher values of the index therefore capture periods of greater economic, geopolitical, and policy tension between the two countries.

---

5. UCT khác EPU và GPR ở đâu?

UCT không phải:

• general Economic Policy Uncertainty;
• general Geopolitical Risk;
• country-specific uncertainty index.

UCT tập trung riêng vào:

$$
U.S.text{–}China bilateral tension
$$

Vì sao quan trọng với Đông Nam Á?

Các nền kinh tế Đông Nam Á:

• có supply-chain integration sâu với China;
• xuất khẩu lớn sang U.S.;
• nhận FDI liên quan đến trade diversion;
• tham gia “China + N” supply-chain reallocation;
• có financing conditions nhạy với global trade shocks.

Do đó UCT có thể tạo spillovers tới financing decisions của Southeast Asian firms dù firms không ở U.S. hoặc China.

Recent IMF research cho thấy ASEAN trade và investment flows phản ứng đáng kể với 2018–2019 U.S.–China tariff increases, củng cố relevance của regional setting.

---

PHẦN D. TẦN SUẤT VÀ MERGE UCT

6. Annual UCT

Cách đơn giản:

$$
UCT_t
$$

được merge với tất cả firms theo year.

Vấn đề lớn

Trong một year:

$$
UCT_t
$$

giống nhau cho:

• Indonesia;
• Malaysia;
• Philippines;
• Singapore;
• Thailand;
• Vietnam;
• và mọi firm.

Do đó:

$$
YearFE
$$

có cùng variation với UCT.

Kết quả:

$$
UCT_t
perp
YearFE
$$

không đúng.

Thực tế UCT bị perfectly collinear với year dummies.

---

7. Design A: Annual UCT Direct-Effect Model

Nếu dùng annual UCT main effect:

$$
Leverage_{i,c,t+1}

beta_1 UCTt
+
Gamma FirmControls{i,t}
+
Delta MacroControls_{c,t}
+
alphai
+
Trend
+
varepsilon{i,c,t}
$$

Có thể dùng

• Firm FE;
• industry controls nếu cần;
• country macro variables;
• deterministic time trend;
• global controls;
• country-specific controls.

Không thể dùng

$$
YearFE
$$

nếu muốn estimate main:

$$
beta_1 UCT_t
$$

Hạn chế

Không có Year FE làm model dễ bị:

• global omitted shocks;
• global business cycle;
• COVID;
• global credit conditions;
• world interest rates.

Do đó Design A nên được coi là baseline đơn giản hoặc descriptive direct-effect design, không phải identification mạnh nhất.

---

8. Design B: Fiscal-Year-Aligned UCT

Nếu UCT có monthly data và Compustat có fiscal-year-end date, có thể xây:

$$
UCT^{FY}_{i,t}

frac{1}{12}
sum{min FY{i,t}}
UCT_m
$$

Ví dụ

Firm A fiscal year end December 2020:

$$
UCT^{FY}_{A,2020}

Average(Jan2020,ldots,Dec2020)
$$

Firm B fiscal year end June 2020:

$$
UCT^{FY}_{B,2020}

Average(Jul2019,ldots,Jun2020)
$$

Do đó hai firms có cùng fiscal-year label có thể có different UCT exposure windows.

Ưu điểm

Có thể tạo:

• cross-firm timing variation;
• better alignment với accounting period.

Có dùng Year FE được không?

Chỉ khi sau fiscal-year alignment vẫn còn đủ variation của UCT trong cùng fiscal-year label.

Sinh viên phải kiểm tra:

Diagnostic	Kết quả
Unique fiscal-year-end months
% December year-end firms
UCT within-year SD
UCT within-year min/max
Year FE absorbs UCT?

Cảnh báo

Nếu hầu hết firms đều December year-end:

• variation vẫn gần như time-only;
• Year FE có thể làm coefficient weakly identified hoặc collinear.

Không mặc định rằng fiscal-year alignment tự động giải quyết vấn đề.

---

9. Design C: UCT × Exposure

Strong design:

$$
CapitalStructure_{i,c,t+1}

beta_1
left(
UCTt
times
Exposure{i/c}
right)
+
FirmFE
+
YearFE
+
Industrytimes YearFE
+
Controls
+
varepsilon
$$

Main UCT

$$
UCT_t
$$

bị Year FE absorb.

Điều này không phải vấn đề nếu parameter quan tâm là:

$$
UCT_ttimes Exposure
$$

Identification

So sánh:

khi U.S.–China tension tăng, firms/countries có exposure cao thay đổi capital structure mạnh hơn firms/countries exposure thấp hay không?

Đây là design nên ưu tiên cho paper mạnh hơn.

---

PHẦN E. MẪU ĐÔNG NAM Á

10. Regional Target Universe

Đông Nam Á hiện có 11 ASEAN member states:

1. Brunei Darussalam
2. Cambodia
3. Indonesia
4. Lao PDR
5. Malaysia
6. Myanmar
7. Philippines
8. Singapore
9. Thailand
10. Timor-Leste
11. Vietnam

Timor-Leste trở thành thành viên ASEAN thứ 11 vào ngày 26 October 2025.

Nhưng

Research sample không được ép phải có đủ 11 countries.

Compustat coverage và listed-firm availability có thể rất thấp hoặc không có ở:

• Brunei;
• Cambodia;
• Lao PDR;
• Myanmar;
• Timor-Leste.

Do đó final sample là data-driven.

---

11. Core Empirical Sample khuyến nghị

Nếu mục tiêu là comparability và sufficient Compustat observations, core sample có thể là:

• Indonesia;
• Malaysia;
• Philippines;
• Singapore;
• Thailand;
• Vietnam.

Ký hiệu:

ASEAN-6

Vì sao?

Nhóm này thường có:

• deeper listed equity markets;
• better firm-level financial coverage;
• longer panel;
• enough observations để estimate firm FE.

Recent ASEAN corporate-finance research cũng thường sử dụng sáu countries này.

Quy tắc

File methodology cuối phải nói:

“The final sample includes X Southeast Asian countries for which both firm-level accounting data and the variables required for the analysis are available.”

Không nói:

“We study all Southeast Asian countries”

nếu actual sample chỉ ASEAN-6.

---

12. Country Selection Table

Sinh viên bắt buộc tạo:

Country	ISO	ASEAN member	Compustat firms	Years available	Keep?	Reason
Brunei	BRN	Yes
Cambodia	KHM	Yes
Indonesia	IDN	Yes
Lao PDR	LAO	Yes
Malaysia	MYS	Yes
Myanmar	MMR	Yes
Philippines	PHL	Yes
Singapore	SGP	Yes
Thailand	THA	Yes
Timor-Leste	TLS	Yes
Vietnam	VNM	Yes

---

PHẦN F. SAMPLE CONSTRUCTION

13. Loại Financial Firms

Theo workflow bài mẫu, loại:

$$
GIND=40
$$

nếu dùng GICS Industry Group.

Nếu dùng SIC:

$$
SIC=6000text{–}6999
$$

Lý do

Financial firms có:

• regulated capital;
• leverage là operating input;
• balance-sheet structure đặc thù;
• prudential regulation;
• funding structure khác industrial firms.

Đoạn viết mẫu

We exclude financial firms because leverage has a structurally different economic meaning in regulated financial institutions, whose balance sheets and capital requirements are shaped by prudential regulation.

---

14. Loại Utilities

Theo workflow bài mẫu:

$$
GIND=55
$$

Nếu dùng SIC:

$$
SIC=4900text{–}4999
$$

Lý do

Utilities có:

• regulated rates;
• distinctive leverage capacity;
• stable cash flows;
• government/regulatory constraints.

Đoạn viết mẫu

We also exclude regulated utilities because their financing capacity and capital-structure choices are heavily influenced by sector-specific regulatory arrangements.

---

15. Total Assets

Require:

$$
AT>0
$$

Lý do

AT là denominator của:

• leverage;
• profitability;
• tangibility;
• cash;
• market-to-book approximations.

---

16. Sales

Require:

$$
SALEgeq0
$$

theo workflow dữ liệu được cung cấp.

Cảnh báo

Không tự đổi thành:

$$
SALE>0
$$

nếu code không dùng rule này.

---

17. Missing Observations

Baseline sample loại observations thiếu:

• DV;
• UCT/exposure;
• firm controls;
• required macro controls;
• identifiers;
• FE variables.

Nếu chạy nhiều DVs

Sample size có thể khác giữa:

• Total Leverage;
• Short-Term Leverage;
• Long-Term Leverage;
• Market Leverage;
• Debt Maturity.

Phải báo N riêng nếu khác nhau.

---

18. Sample Selection Table

Sample construction step	Observations removed	Observations remaining
Initial Southeast Asia Compustat sample
Restrict to selected countries
Exclude GIND 40
Exclude GIND 55
Require AT > 0
Require SALE ≥ 0
Remove missing debt variables
Merge UCT
Merge macro variables
Remove missing controls
Final baseline sample

Ngoài observations, báo:

• firms;
• countries;
• industries;
• years.

---

PHẦN G. OUTLIERS VÀ WINSORIZATION

19. Winsorization Rule

Theo workflow bài mẫu:

• Dependent variables được winsorize;
• Firm-level controls được winsorize;
• mức:

$$
1^{st}text{ and }99^{th} percentiles
$$

Không winsorize

• UCT;
• macro variables;
• identifiers;
• country variables;
• year;
• dummies.

Tại sao macro không winsorize?

Macro variables phản ánh genuine country-level variation.

Không nên áp firm-level outlier treatment một cách máy móc.

Đoạn viết mẫu

To mitigate the influence of extreme firm-level accounting observations, we winsorize the dependent variables and continuous firm-level controls at the 1st and 99th percentiles. We do not winsorize the U.S.–China Tension Index or macroeconomic variables because their extreme values may represent economically meaningful aggregate shocks.

---

PHẦN H. CAPITAL STRUCTURE MEASURES

20. Main Measure: Book Leverage

Main dependent variable:

$$
BookLeverage_{i,t}

frac{
DLC{i,t}+DLTT{i,t}
}{
AT_{i,t}
}
$$

Trong đó:

• $DLC$ = Debt in Current Liabilities / Short-Term Debt;
• $DLTT$ = Long-Term Debt;
• $AT$ = Total Assets.

Interpretation

$$
Higher BookLeverage

Greater Use of Interesttext{-}Bearing Debt
$$

Tại sao dùng debt thay vì Total Liabilities?

Capital structure focuses on financing debt/equity mix.

Total liabilities có thể gồm:

• accounts payable;
• tax liabilities;
• operating liabilities

không phải financing debt.

Đoạn viết mẫu

Our primary capital-structure measure is book leverage, defined as short-term plus long-term interest-bearing debt divided by total assets. This measure captures the firm’s reliance on debt financing while excluding non-debt operating liabilities.

---

21. Short-Term Leverage

$$
STLeverage_{i,t}

frac{DLC{i,t}}{AT{i,t}}
$$

Tại sao đặc biệt quan trọng?

U.S.–China tension có thể làm tăng:

• rollover risk;
• refinancing uncertainty;
• short-term credit tightening.

Firms có thể giảm short-term borrowing trước tiên.

Direct empirical precedent

Bui Tien Thinh et al. (2026) nghiên cứu Vietnamese firms và tìm thấy U.S.–China tensions làm giảm:

• overall leverage;
• short-term leverage.

Đoạn viết mẫu

We separately examine short-term leverage because short-maturity borrowing exposes firms to rollover and refinancing risk. If U.S.–China tensions tighten credit conditions or increase uncertainty about future cash flows, firms may respond particularly strongly by reducing short-term debt.

---

22. Long-Term Leverage

$$
LTLeverage_{i,t}

frac{DLTT{i,t}}{AT{i,t}}
$$

Mục tiêu

Kiểm tra UCT ảnh hưởng:

• total debt level;
• debt maturity composition;
• long-term financing separately.

Interpretation

Nếu UCT giảm short-term leverage nhưng long-term leverage không đổi:

firms may be reducing refinancing exposure rather than deleveraging uniformly.

Nếu long-term leverage giảm mạnh:

uncertainty may reduce access to long-horizon external financing.

---

23. Debt Maturity

Một common measure:

$$
DebtMaturity_{i,t}

frac{DLTT{i,t}}
{DLC{i,t}+DLTT_{i,t}}
$$

với condition:

$$
DLC+DLTT>0
$$

Interpretation

$$
Higher DebtMaturity

Greater Share of Longtext{-}Term Debt
$$

Cảnh báo

Zero-debt firms không có defined debt-maturity ratio.

Phải:

• exclude khỏi debt-maturity regression;
• hoặc define theo exact methodology.

Không tự đặt bằng 0.

---

24. Market Leverage

Một common approximation:

$$
MarketLeverage_{i,t}

frac{
DLC+DLTT
}{
DLC+DLTT+MarketValueEquity
}
$$

với:

$$
MarketValueEquity

PRCC_Ftimes CSHO
$$

nếu exact Compustat construction phù hợp.

Quy tắc

Phải khớp code.

Không dùng một Tobin/market-value formula từ paper khác nếu data construction thực tế khác.

---

25. Capital Structure Variable Table

Variable	Role	Formula	Higher value means	Data source	Reference
Book Leverage	Main DV	(DLC+DLTT)/AT	More debt financing	Compustat	Capital-structure literature
ST Leverage	Alternative DV	DLC/AT	More short-term debt	Compustat	Bui et al. (2026)
LT Leverage	Alternative DV	DLTT/AT	More long-term debt	Compustat
Debt Maturity	Alternative DV	DLTT/(DLC+DLTT)	Longer debt maturity	Compustat	Debt-maturity literature
Market Leverage	Alternative DV	Debt/(Debt+MVE)	More market leverage	Compustat

---

PHẦN I. CAPITAL STRUCTURE THEORY

26. Modigliani and Miller

Modigliani and Miller (1958) provide the foundational irrelevance benchmark under frictionless markets.

Real-world capital structure becomes relevant once introducing:

• taxes;
• bankruptcy costs;
• information asymmetry;
• agency problems;
• financing frictions.

---

27. Trade-Off Theory

Firms trade off:

• tax benefits of debt;
• expected financial-distress costs.

Under high UCT:

$$
CashFlowRiskuparrow
$$

$$
DefaultRiskuparrow
$$

$$
ExpectedDistressCostuparrow
$$

which may imply:

$$
OptimalLeveragedownarrow
$$

Prediction

$$
UCTuparrow
rightarrow
Leveragedownarrow
$$

---

28. Pecking-Order Theory

Myers (1984) and Myers and Majluf (1984) emphasize information asymmetry.

Firms prefer:

1. internal funds;
2. debt;
3. equity.

UCT can create ambiguous predictions.

Channel A

If uncertainty reduces investment and firms preserve internal funds:

$$
DebtDemanddownarrow
$$

Channel B

If cash flows fall and internal financing becomes insufficient:

$$
DebtDemanduparrow
$$

unless credit supply simultaneously contracts.

Therefore the final sign remains empirical.

---

29. Financial Flexibility Channel

High uncertainty increases value of:

• unused debt capacity;
• cash buffers;
• lower refinancing exposure.

Firms may deleverage to maintain:

$$
FinancialFlexibility
$$

for future shocks.

This channel predicts:

$$
UCTuparrow
rightarrow
Leveragedownarrow
$$

especially for:

• financially constrained firms;
• firms with low cash;
• firms highly exposed to trade shocks.

---

30. Credit Supply Channel

Banks and bond investors may respond to UCT by:

• tightening lending standards;
• charging higher spreads;
• shortening maturity;
• reducing credit quantity.

This can mechanically reduce observed leverage.

Important

Observed leverage reflects both:

• firm debt demand;
• creditor debt supply.

Paper should not interpret leverage reduction as purely managerial preference.

---

PHẦN J. CONTROL VARIABLES

31. Core Firm Controls

Canonical capital-structure literature identifies important determinants including:

• Size;
• Tangibility;
• Profitability;
• Market-to-Book / Growth Opportunities.

Rajan and Zingales (1995) provide major international evidence.

Frank and Goyal (2009) identify robust leverage determinants.

---

32. Tangibility

$$
Tangibility

frac{PPENT}{AT}
$$

Prediction

Tangible assets provide collateral.

Expected:

$$
+
$$

with leverage.

---

33. Profitability

Example:

$$
Profitability

frac{IB}{AT}
$$

or exact code measure.

Pecking-order prediction

Profitable firms generate more internal funds:

$$
Profitabilityuparrow
rightarrow
DebtNeeddownarrow
$$

Expected:

$$

$$

often observed empirically.

---

34. Size

$$
Size

ln(AT)
$$

Larger firms may have:

• lower default risk;
• better capital-market access;
• greater diversification.

Expected leverage sign often:

$$
+
$$

but must follow empirical literature.

---

35. Market-to-Book

Exact formula must follow code.

Market-to-Book captures:

• growth opportunities;
• valuation.

Under many trade-off interpretations:

$$
Higher growth opportunities
rightarrow
Lower leverage
$$

due to distress/underinvestment concerns.

---

36. Cash Holdings

Optional:

$$
Cash

frac{CHE}{AT}
$$

Rationale

Cash provides internal financing and financial flexibility.

Can reduce immediate debt need.

---

37. Sales Growth

Optional firm control.

Caution

Sales growth can be affected by UCT.

If UCT affects sales growth and sales growth affects leverage, controlling for Sales Growth may block a transmission channel.

Therefore classify it carefully as:

• baseline control;
• expanded control;
• or mechanism variable.

---

38. Core Macro Controls

Because annual UCT direct-effect Design A cannot use Year FE, macro controls become especially important.

Candidate variables:

• GDP Growth;
• Inflation;
• Interest Rate;
• Exchange Rate changes;
• domestic credit conditions;
• financial development.

User workflow rule

Macro variables:

do not winsorize.

---

39. GDP Growth

$$
GDPGrowth_{c,t}
$$

Captures country business cycle.

---

40. Inflation

$$
Inflation_{c,t}
$$

Affects:

• real debt burden;
• nominal interest rates;
• financing conditions.

---

41. Global Controls for Design A

Because UCT is a global time-series variable, consider controls such as:

• Global EPU;
• GPR;
• VIX / global financial conditions;
• world GDP growth;
• U.S. interest rates.

Mục tiêu

Reduce risk that UCT merely proxies for:

• global recession;
• general geopolitical stress;
• global financial tightening.

Caution

Do not load many highly correlated global indices into one model without:

• correlation analysis;
• multicollinearity check;
• coefficient stability analysis.

---

PHẦN K. TIMING

42. Main Timing

Recommended:

$$
UCTt
rightarrow
CapitalStructure{i,c,t+1}
$$

Rationale

Capital-structure adjustment may take time because:

• debt contracts mature;
• firms refinance;
• banks revise lending policies;
• managers alter financing plans.

Đoạn viết mẫu

We relate U.S.–China tension in year $t$ to corporate capital structure in year $t+1$, allowing the external tension shock to precede subsequent financing decisions. This timing provides a clearer temporal ordering and reflects the fact that leverage adjustments may occur gradually as debt contracts mature and firms refinance.

Caution

Lead Y does not solve all endogeneity.

---

43. Alternative Timing

Robustness:

$$
UCT_{t-1}
rightarrow
Leverage_t
$$

$$
UCT_t
rightarrow
Leverage_t
$$

$$
UCTt
rightarrow
Leverage{t+2}
$$

---

PHẦN L. BASELINE EMPIRICAL DESIGNS

44. Design A: Annual Direct Effect

$$
CapitalStructure_{i,c,t+1}

beta_1 UCTt
+
Gamma FirmControls{i,t}
+
Delta MacroControls_{c,t}
+
alphai
+
Trend
+
varepsilon{i,c,t}
$$

Include

• Firm FE;
• country macro controls;
• global controls;
• optional time trend.

Exclude

• Year FE.

Limitation

Potential omitted common time shocks remain.

This limitation must be acknowledged.

---

45. Design B: Fiscal-Year-Aligned UCT

If:

$$
UCT^{FY}_{i,t}
$$

has sufficient within-year variation:

$$
CapitalStructure_{i,c,t+1}

beta1 UCT^{FY}{i,t}
+
Controls
+
FirmFE
+
YearFE
+
varepsilon
$$

Before use

Report:

• fiscal-year-end distribution;
• within-year UCT variation;
• collinearity diagnostic.

---

46. Design C: Exposure-Based UCT

Country-level exposure

Example:

$$
TradeExposure_c

frac{
TradeWithUS_c+TradeWithChina_c
}{
GDP_c
}
$$

Then:

$$
Leverage_{i,c,t+1}

beta_1
left(
UCT_t
times
TradeExposure_c
right)
+
FirmFE
+
YearFE
+
Industrytimes YearFE
+
Controls
+
varepsilon
$$

Main advantage

Year FE absorb global UCT main effect and all common year shocks.

Identification comes from:

countries with different exposure respond differently to the same UCT shock.

Caution

If Exposure is time-varying country-year:

interaction itself becomes country-year.

Do not include:

$$
Countrytimes YearFE
$$

because it would absorb the interaction.

---

47. Firm-Level Exposure Design

If firm-level export/supply-chain exposure is available:

$$
Exposure_{i,t}
$$

then:

$$
Leverage_{i,c,t+1}

beta_1
left(
UCTttimes Exposure{i,t}
right)
+
FirmFE
+
Countrytimes YearFE
+
Industrytimes YearFE
+
varepsilon
$$

This is stronger

Country × Year FE absorb:

• all country-year macro shocks;
• country-level credit conditions;
• domestic regulation;
• UCT main effect.

Interaction remains identified because exposure differs across firms within country-year.

Potential firm exposure proxies

• export intensity;
• sales to U.S.;
• sales to China;
• supplier/customer links;
• foreign revenue share;
• industry-level tariff exposure;
• China/U.S. supply-chain exposure.

Compustat alone may not contain all of these reliably.

Do not claim firm-level trade exposure unless source exists.

---

PHẦN M. FIXED EFFECTS

48. Firm Fixed Effects

Useful across all designs.

Absorb:

• persistent business model;
• managerial culture;
• long-run financing preference;
• time-invariant industry/country traits at firm level.

Sample paragraph

We include firm fixed effects to absorb persistent unobserved firm characteristics that may influence financing choices, such as long-run business models, managerial risk preferences, and stable access to external finance.

---

49. Year Fixed Effects

Design A annual UCT main effect

Cannot use.

Design B fiscal-year-aligned UCT

May use if UCT retains within-year variation.

Design C UCT × Exposure

Should use.

Main UCT is absorbed, interaction remains.

Critical rule

Fixed-effects choice depends on identification design.

There is no universal:

“Firm FE + Year FE is always best.”

---

50. Industry × Year Fixed Effects

Useful in Design B/C.

Absorb:

• sector demand shocks;
• industry credit cycles;
• commodity-price effects;
• sector-specific trade shocks common across countries.

In Design A

Can potentially use industry-year FE only if UCT remains identified across industries?

No.

Because UCT_t is common across all firms in year t, a full set of Industry × Year FE also contains year-specific intercepts for each industry and will absorb a pure time-only UCT main effect.

Therefore:

Do not use Industry × Year FE in annual direct-effect Design A.

This point must be checked carefully.

---

51. Country Fixed Effects

With Firm FE and firms not changing countries:

country FE are absorbed by firm FE.

Do not add country FE mechanically.

---

PHẦN N. VCE / STANDARD ERRORS

52. Why the VCE problem is different here

Previous country-year indices such as:

$$
TUI_{c,t}
$$

or:

$$
IQ_{c,t}
$$

have cross-country variation.

UCT annual series:

$$
UCT_t
$$

has only time variation.

Therefore clustering only by:

• firm;
• or country

does not directly address the common-year nature of the regressor.

---

53. Design A VCE

For annual UCT direct-effect regressions, consider:

Two-way clustering

$$
Firm times Year
$$

or:

$$
Country times Year
$$

depending residual-dependence structure.

Why cluster by Year?

All firms are exposed to same:

$$
UCT_t
$$

in each year.

Common shocks can create cross-sectional residual correlation.

Why also cluster by Firm?

Panel residuals may be serially correlated within firms.

Method source

Cameron, Gelbach, and Miller (2011) provide multiway cluster-robust inference.

Caution

Number of years may be small.

Report:

$$
N_{year clusters}
$$

and consider inference sensitivity.

---

54. Design B VCE

If fiscal-year-aligned UCT varies by firm timing:

baseline can consider:

• firm clustering;
• two-way firm and year;
• alternative country/year clustering.

Choice must follow actual residual structure and remaining UCT variation.

---

55. Design C Country-Exposure VCE

If treatment:

$$
UCT_ttimes Exposure_c
$$

varies at country-year level:

consider:

• cluster country;
• two-way country and year.

Few countries

If core sample = ASEAN-6:

$$
N_{country}=6
$$

Conventional country-clustered SE is highly unreliable.

Therefore country-level wild cluster methods with only six clusters may also have limitations.

Important

With ASEAN-6, inference must be designed carefully.

Possible robustness:

• randomization/permutation-style inference if defensible;
• wild bootstrap with caution;
• Driscoll-Kraay as sensitivity;
• leave-one-country-out;
• country-by-country estimates;
• event/exposure design;
• report coefficient stability rather than relying only on asymptotic p-values.

No method magically solves having six independent country clusters.

---

56. Design C Firm-Level Exposure VCE

If exposure varies at firm level:

can consider:

• cluster firm;
• two-way firm and year;
• country-year shock absorbed by Country × Year FE.

This is statistically more attractive than a pure country-exposure design if credible firm-level exposure data exist.

---

PHẦN O. HETEROGENEITY

57. Financial Constraints

Direct Vietnam evidence finds UCT effects on leverage are stronger for financially constrained firms.

Possible proxies:

• Size;
• Sales;
• SA Index;
• WW Index;
• dividend status.

Prefer continuous interaction where possible

rather than arbitrary median splits.

---

58. Cash Holdings

$$
UCTtimes LowCash
$$

If high UCT leads firms to preserve flexibility, low-cash firms may respond more strongly.

---

59. Export/Trade Exposure

Strong candidate.

Test:

$$
UCT
times
HighTradeExposure
$$

Could be:

• country exposure;
• industry exposure;
• firm exposure.

---

60. China vs U.S. Exposure

Separate:

$$
UCTtimes ChinaExposure
$$

and:

$$
UCTtimes USExposure
$$

because ASEAN countries/firms may gain from:

• trade diversion;
• China+1 relocation;

while suffering from:

• supply-chain disruptions;
• tariff uncertainty.

Effects need not be symmetric.

---

61. Short-Term Debt Dependence

Firms with historically high short-term debt may react more strongly due to refinancing risk.

Test:

$$
UCT
times
HighSTDebt
$$

---

62. Country Financial Development

UCT effect may differ in:

• bank-based systems;
• deeper capital markets;
• stronger financial institutions.

Potential country variables:

• Domestic Credit/GDP;
• stock-market development;
• bank concentration;
• institutional quality.

---

PHẦN P. ASEAN / SOUTHEAST ASIA CONTEXT

63. Why Southeast Asia?

The region is useful because it sits economically between the U.S. and China.

Potential channels:

Trade Diversion

U.S.–China tariffs may redirect production and exports toward ASEAN.

This can improve:

• investment opportunities;
• cash flows;
• credit demand.

Supply-Chain Disruption

Firms dependent on Chinese inputs or U.S. demand may face:

• higher costs;
• uncertain demand;
• working-capital pressure.

FDI Reallocation

Manufacturing and supply chains may relocate into Southeast Asia.

Financial Spillovers

Global risk and trade uncertainty can change:

• bank lending;
• bond financing;
• exchange rates;
• investor risk appetite.

Therefore

The sign of UCT on leverage may vary across:

• countries;
• industries;
• exposure types.

---

PHẦN Q. DIRECT EMPIRICAL PRECEDENT

64. Bui Tien Thinh et al. (2026)

Paper:

Corporate leverage and U.S.-China tensions: evidence from Vietnam

Journal:

Cogent Economics & Finance, 14(1), Article 2613610.

DOI:

10.1080/23322039.2026.2613610

Sample:

• Vietnamese listed nonfinancial firms;
• 2017–2023;
• 8,315 observations;
• 1,301 firms.

Main outcomes:

• overall leverage;
• short-term leverage.

Main finding:

higher U.S.–China tensions are associated with:

• lower overall leverage;
• lower short-term leverage.

Additional evidence:

effects stronger for:

• financially constrained firms;
• some industries;
• COVID period.

Methods include:

• baseline regressions;
• 2SLS;
• dynamic GMM robustness.

Why this paper is critical

This is the closest empirical reference to the current Southeast Asia study.

The new paper should not merely replicate it.

The contribution must come from:

• multi-country Southeast Asia sample;
• cross-country heterogeneity;
• debt maturity;
• trade exposure;
• institutional/financial-system differences;
• stronger exposure-based identification.

---

PHẦN R. CAPITAL STRUCTURE REFERENCES

65. Foundational Theory

Modigliani, F., & Miller, M. H. (1958).
The Cost of Capital, Corporation Finance and the Theory of Investment.
American Economic Review, 48(3), 261–297.

Jensen, M. C., & Meckling, W. H. (1976).
Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure.
Journal of Financial Economics, 3(4), 305–360.
DOI: 10.1016/0304-405X(76)90026-X

Myers, S. C. (1984).
The Capital Structure Puzzle.
The Journal of Finance, 39(3), 575–592.
DOI: 10.1111/j.1540-6261.1984.tb03646.x

Myers, S. C., & Majluf, N. S. (1984).
Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have.
Journal of Financial Economics, 13(2), 187–221.

---

66. Empirical Capital Structure

Rajan, R. G., & Zingales, L. (1995).
What Do We Know about Capital Structure? Some Evidence from International Data.
The Journal of Finance, 50(5), 1421–1460.

Frank, M. Z., & Goyal, V. K. (2009).
Capital Structure Decisions: Which Factors Are Reliably Important?
Financial Management, 38(1), 1–37.

Lemmon, M. L., Roberts, M. R., & Zender, J. F. (2008).
Back to the Beginning: Persistence and the Cross-Section of Corporate Capital Structure.
The Journal of Finance, 63(4), 1575–1608.

---

PHẦN S. ASEAN CAPITAL-STRUCTURE REFERENCES

67. ASEAN Regional Evidence

Recent and regional studies can support:

• use of leverage measures;
• country setting;
• differences across financing systems.

Examples include research on:

• Malaysia, Singapore, and Thailand capital-structure determinants;
• family ownership and capital structure across ASEAN;
• speed of adjustment toward target leverage among ASEAN firms;
• debt financing and firm performance in ASEAN-6.

Student requirement

Do not use these papers only to say:

“ASEAN is interesting.”

Extract:

• sample countries;
• leverage measure;
• controls;
• FE;
• VCE;
• country-level variables.

---

PHẦN T. METHOD DECISION MATRIX

68. Bảng bắt buộc

Methodological decision	What the paper does	Why?	Supporting reference	Verified?
Firm database	Compustat Global	International firms
Region	Southeast Asia	U.S.–China spillover setting
Core sample	ASEAN-6 / actual countries	Data availability	ASEAN studies
UCT source	Rogers et al. (2024)	Bilateral tension	Rogers et al.
UCT frequency
Fiscal-year alignment
Exclude GIND 40	Yes	Financial regulation	Capital-structure literature
Exclude GIND 55	Yes	Utility regulation	Capital-structure literature
AT > 0	Yes	Valid ratios
SALE ≥ 0	Yes	Valid data
Winsor DV	1st–99th	Outliers	Empirical literature
Winsor firm controls	1st–99th	Outliers	Empirical literature
Winsor macro	No	Preserve macro variation
Winsor UCT	No	Preserve tension shocks
Main DV	Book Leverage	Debt financing	Capital-structure literature
ST Leverage	Alternative	Refinancing channel	Bui et al. (2026)
LT Leverage	Alternative	Long-term financing
Debt Maturity	Alternative	Maturity structure	Debt-maturity literature
Market Leverage	Alternative	Market-value structure	Capital-structure literature
Timing	UCT_t → Y_t+1	Temporal ordering
Design A Year FE	No	UCT absorbed	Identification logic
Design B Year FE	Conditional	Depends on FY variation
Design C Year FE	Yes	Global shocks	Exposure design
Firm FE	Yes	Firm heterogeneity
Industry-year FE	Only Design B/C	Sector shocks
Country-year FE	Only with firm-level exposure interaction	Strong control	Identification logic
VCE Design A	Two-way firm/year or appropriate alternative	Time-only X	Cameron et al.
VCE Design C	Country/year or firm/year depending exposure	Treatment level	Cameron et al.
Few-cluster checks	Yes	ASEAN sample small	Cluster literature

---

PHẦN U. VARIABLE DEFINITION TABLE

69. Bảng bắt buộc

Variable	Role	Definition / Formula	Higher value means	Data source	Reference
Book Leverage	Main DV	(DLC+DLTT)/AT	More debt	Compustat	Rajan & Zingales; Frank & Goyal
ST Leverage	Alt DV	DLC/AT	More short-term debt	Compustat	Bui et al. (2026)
LT Leverage	Alt DV	DLTT/AT	More long-term debt	Compustat
Debt Maturity	Alt DV	DLTT/(DLC+DLTT)	Longer maturity	Compustat
Market Leverage	Alt DV	Debt/(Debt+MVE)	More market leverage	Compustat
UCT	Main IV	U.S.–China Tension Index	Higher bilateral tension	PolicyUncertainty.com	Rogers et al. (2024)
Tangibility	Control	PPENT/AT	More collateral	Compustat	Rajan & Zingales
Profitability	Control	Exact code	More profitable	Compustat	Frank & Goyal
Size	Control	ln(AT)	Larger firm	Compustat
Market-to-Book	Control	Exact code	More growth opportunities	Compustat
Cash	Control	CHE/AT	More liquidity	Compustat
GDP Growth	Macro		Stronger growth	WDI
Inflation	Macro		Higher inflation	WDI
Trade Exposure	Exposure	Exact definition	More U.S./China exposure	External trade data

---

PHẦN V. IDENTIFICATION DIAGNOSTIC TABLE

70. Bảng này bắt buộc vì UCT là time-series index

Diagnostic	Result
UCT frequency
UCT varies by country?	No / Verify
UCT varies within year?
Fiscal-year aligned?
Number of unique UCT values per year
Correlation with year dummies / collinearity test
Design selected	A / B / C
Year FE allowed?
Industry × Year FE allowed?
Country × Year FE allowed?
Exposure level	None / Country / Industry / Firm
Main independent variation used for identification

Rule

Sinh viên chưa hoàn thành bảng này thì chưa được chạy baseline regression.

---

PHẦN W. COUNTRY SAMPLE TABLE

71. Bảng bắt buộc

Country	ISO	Initial firms	Final firms	Firm-years	First year	Last year	Keep?
Brunei	BRN
Cambodia	KHM
Indonesia	IDN
Lao PDR	LAO
Malaysia	MYS
Myanmar	MMR
Philippines	PHL
Singapore	SGP
Thailand	THA
Timor-Leste	TLS
Vietnam	VNM

---

PHẦN X. MODEL REFERENCE MATRIX

72. Bảng bắt buộc

Study	Sample	Main X	Capital-structure Y	Controls	FE/Model	VCE	Bài này học gì?
Rogers et al. (2024)	U.S./global evidence	UCT	Investment/supply chains				UCT construction
Bui et al. (2026)	Vietnam firms	UCT	Leverage, ST leverage		OLS/2SLS/GMM		Closest direct precedent
Rajan & Zingales (1995)	International firms	Firm determinants	Leverage				Controls/international design
Frank & Goyal (2009)	Firms	Determinants	Leverage				Core leverage determinants
ASEAN capital structure paper 1	ASEAN						Regional controls
ASEAN capital structure paper 2	ASEAN						Regional heterogeneity
Debt maturity paper			Debt maturity				Maturity measure
Moulton/Cameron method		Aggregate X					Inference

Tối thiểu:

8 model/method references.

---

PHẦN Y. ĐOẠN MẪU DATA AND METHODOLOGY

73. Sample paragraph: Regional sample

We construct a firm-year panel of publicly listed companies in Southeast Asia using Compustat Global. Our target universe consists of Southeast Asian economies for which sufficiently complete firm-level accounting data are available. The core estimation sample focuses on Indonesia, Malaysia, the Philippines, Singapore, Thailand, and Vietnam, while additional countries are retained when the data satisfy the required coverage and variable-availability criteria. We exclude financial firms and regulated utilities, require positive total assets and valid nonnegative sales, and remove observations with missing variables required for the baseline regressions. Continuous firm-level dependent variables and controls are winsorized at the 1st and 99th percentiles, whereas aggregate UCT and macroeconomic variables are not winsorized.

---

74. Sample paragraph: UCT

Our main explanatory variable is the U.S.–China Tension Index developed by Rogers et al. (2024) and distributed through the Economic Policy Uncertainty website. The index is constructed from leading U.S. newspapers and measures the share of articles that jointly reference the United States and China, contentious bilateral issues, and tension-related language. Higher values indicate more intense U.S.–China tension. Because UCT is a global time-series measure rather than a country-specific index, its variation structure has direct implications for our fixed-effects design.

---

75. Sample paragraph: Capital Structure

Our primary capital-structure measure is book leverage, defined as the sum of short-term and long-term interest-bearing debt divided by total assets. We additionally examine short-term leverage and long-term leverage to determine whether tension affects firms’ refinancing exposure and long-horizon borrowing differently. Debt maturity, defined as long-term debt relative to total interest-bearing debt, provides a complementary measure of financing structure, while market leverage is used as an alternative valuation-based specification.

---

76. Sample paragraph: Direct-effect identification

A key feature of the UCT index is that the same aggregate tension value applies to all firms at a given point in time. Consequently, when annual UCT is used, a full set of year fixed effects would absorb the index and prevent estimation of its direct effect. Our annual direct-effect specification therefore includes firm fixed effects and a set of firm-level and macroeconomic controls but does not include year fixed effects. Because this design remains vulnerable to omitted common time shocks, we complement it with specifications based on fiscal-year-aligned UCT and exposure interactions that allow stronger time fixed effects.

---

77. Sample paragraph: Exposure-based identification

Our stronger specification interacts U.S.–China tension with firms’ or countries’ economic exposure to the United States and China. This design allows us to include year fixed effects, which absorb aggregate global shocks as well as the main effect of UCT. The interaction coefficient then captures whether firms with greater exposure adjust their capital structures more strongly when U.S.–China tensions rise. Where credible firm-level exposure measures are available, we further include country-by-year fixed effects so that identification comes from differences in exposure across firms operating in the same country and year.

---

78. Sample paragraph: VCE

The appropriate inference procedure depends on the source of identifying variation. In annual direct-effect models, UCT varies primarily over time and is shared by all firms, so standard errors must account for common year shocks in addition to serial dependence within firms. We therefore consider multiway clustering by firm and year and report the number of year clusters. In exposure-based designs where the treatment varies at the country-year level, we examine country- and year-level clustering and explicitly assess the limitations created by the relatively small number of Southeast Asian countries.

---

PHẦN Z. CORE REFERENCES ĐƯỢC CUNG CẤP

79. U.S.–China Tension

Rogers, J. H., Sun, B., & Sun, T. (2024).
U.S.–China Tension.
Working paper / SSRN.
DOI: 10.2139/ssrn.4815838

Data available through:

Economic Policy Uncertainty – U.S.–China Tension Index

---

80. Closest Direct Capital-Structure Paper

Bui Tien Thinh, Le Thi My Nuong, Ha Thi Thuy, & Dao Nhat Minh. (2026).
Corporate Leverage and U.S.-China Tensions: Evidence from Vietnam.
Cogent Economics & Finance, 14(1), 2613610.
DOI: 10.1080/23322039.2026.2613610

Key relevance:

• UCT;
• leverage;
• short-term leverage;
• Vietnam;
• financial constraints;
• industry heterogeneity;
• 2SLS;
• GMM.

---

81. UCT Spillover / Economic Effects

Rogers, J. H., Sun, B., & Sun, T. (2024).
The original UCT research documents links between U.S.–China tension and corporate investment, particularly among more China-exposed firms.

Attílio, L. A. (2026).
U.S.–China Tensions and Macroeconomic Fluctuations.
The Manchester School.
DOI: 10.1111/manc.70067

Useful for:

• international spillovers;
• macro transmission.

Luo, C. (2026).
U.S.–China Trade Tensions and Their Supply Chain Relationships: Evidence From U.S. Listed Firms.
Review of International Economics, 34(3), 579–596.
DOI: 10.1111/roie.70039

Useful for:

• supply-chain reconfiguration;
• China+N / de-risking channels.

---

82. ASEAN Trade and Fragmentation

Donato, G., & Kitsios, E. (2026).
ASEAN’s Trade and Investment in a Fragmented World.
IMF Working Paper.

Useful for:

• 2018–2019 U.S.–China tariff spillovers;
• ASEAN trade diversion;
• ASEAN investment reallocation.

---

83. Capital Structure Foundations

Modigliani, F., & Miller, M. H. (1958).
The Cost of Capital, Corporation Finance and the Theory of Investment.
American Economic Review, 48(3), 261–297.

Jensen, M. C., & Meckling, W. H. (1976).
Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure.
Journal of Financial Economics, 3(4), 305–360.

Myers, S. C. (1984).
The Capital Structure Puzzle.
The Journal of Finance, 39(3), 575–592.

Myers, S. C., & Majluf, N. S. (1984).
Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have.
Journal of Financial Economics, 13(2), 187–221.

---

84. Empirical Capital Structure

Rajan, R. G., & Zingales, L. (1995).
What Do We Know about Capital Structure? Some Evidence from International Data.
The Journal of Finance, 50(5), 1421–1460.

Frank, M. Z., & Goyal, V. K. (2009).
Capital Structure Decisions: Which Factors Are Reliably Important?
Financial Management, 38(1), 1–37.

Lemmon, M. L., Roberts, M. R., & Zender, J. F. (2008).
Back to the Beginning: Persistence and the Cross-Section of Corporate Capital Structure.
The Journal of Finance, 63(4), 1575–1608.

---

85. ASEAN Capital Structure

Students must identify at least three regional empirical studies.

Suggested lines include:

• determinants of capital structure in Malaysia, Singapore, and Thailand;
• family ownership and capital structure across ASEAN countries;
• speed of adjustment toward target capital structure among ASEAN firms;
• debt financing patterns across ASEAN-6.

The purpose is not merely to cite ASEAN papers but to extract:

• leverage definitions;
• country sample;
• firm controls;
• macro controls;
• estimation method;
• FE;
• VCE.

---

86. Clustered Inference

Moulton, B. R. (1990).
An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Units.
The Review of Economics and Statistics, 72(2), 334–338.
DOI: 10.2307/2109724

Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2011).
Robust Inference With Multiway Clustering.
Journal of Business & Economic Statistics, 29(2), 238–249.
DOI: 10.1198/jbes.2010.07136

Cameron, A. C., & Miller, D. L. (2015).
A Practitioner’s Guide to Cluster-Robust Inference.
Journal of Human Resources, 50(2), 317–372.
DOI: 10.3368/jhr.50.2.317

---

PHẦN AA. TÀI LIỆU SINH VIÊN PHẢI TỰ TÌM THÊM

87. UCT → Corporate Decisions

Tối thiểu:

3 papers

gồm:

• leverage;
• investment;
• supply chain;
• financial markets.

---

88. Uncertainty → Capital Structure

Tối thiểu:

4 papers

về:

• EPU and leverage;
• geopolitical risk and leverage;
• financial flexibility;
• short-term debt / rollover risk.

---

89. ASEAN Capital Structure

Tối thiểu:

3 papers

ưu tiên multi-country Southeast Asian samples.

---

90. Debt Maturity

Tối thiểu:

2 papers

về:

• short-term debt;
• long-term debt;
• maturity structure;
• rollover risk.

---

91. Exposure-Based Design

Tối thiểu:

2 papers

dùng:

• trade exposure;
• tariff exposure;
• China exposure;
• U.S. exposure;
• supply-chain exposure.

---

92. Controls

Mỗi baseline firm control phải có empirical reference.

Macro controls cũng phải có rationale.

Tổng số Methodology references

Khuyến nghị:

18–25 papers/sources

trước khi chuyển sang Theory.

---

PHẦN AB. QUY TẮC SỬ DỤNG AI

93. AI được phép hỗ trợ

AI có thể:

• giải thích UCT;
• giải thích leverage formulas;
• giải thích debt maturity;
• giải thích trade-off / pecking-order;
• giải thích FE collinearity;
• kiểm tra whether Year FE absorb UCT;
• giải thích clustering;
• hỗ trợ literature-search keywords;
• kiểm tra consistency;
• hỗ trợ viết methodology.

---

94. AI không được thay thế kiểm chứng

Không được:

• tự tạo UCT frequency;
• tự tạo UCT values;
• tự đoán countries;
• tự tạo Compustat sample N;
• tự tạo formulas không khớp code;
• tự tạo exposure variables;
• tự nói Year FE dùng được nếu chưa kiểm tra variation;
• tự tạo p-values;
• tự tạo DOI;
• tự tạo citations.

Mọi thông tin phải kiểm tra bằng:

• original UCT paper;
• PolicyUncertainty.com;
• Compustat data dictionary;
• original papers;
• World Bank/IMF/official trade data;
• actual code and regression output.

---

PHẦN AC. CHECKLIST CUỐI BƯỚC 1

Trước khi chuyển sang Bước 2:

☐ Đã đọc Rogers et al. (2024).
☐ Đã xác định exact UCT file.
☐ Đã xác định UCT frequency.
☐ Đã xác định UCT là global time-series variable.
☐ Đã hiểu annual UCT bị Year FE absorb.
☐ Đã chọn Design A, B hoặc C.
☐ Nếu Design A: không dùng Year FE.
☐ Nếu Design A: đã xác định global/time controls.
☐ Nếu Design A: đã xác định VCE xử lý year shocks.
☐ Nếu Design B: đã xây fiscal-year-aligned UCT.
☐ Nếu Design B: đã kiểm tra within-year UCT variation.
☐ Nếu Design B: đã kiểm tra Year FE collinearity.
☐ Nếu Design C: đã xác định exposure variable.
☐ Nếu Design C: đã xác định exposure ở country, industry hay firm level.
☐ Nếu Design C: đã xác định FE phù hợp.
☐ Đã xác định Southeast Asia country universe.
☐ Đã tạo Country Selection Table.
☐ Đã xác định actual countries trong sample.
☐ Nếu dùng ASEAN-6: đã ghi rõ sáu countries.
☐ Đã xác định lý do loại GIND 40.
☐ Đã xác định lý do loại GIND 55.
☐ Đã kiểm tra AT > 0.
☐ Đã kiểm tra SALE ≥ 0.
☐ Đã xác định missing-data treatment.
☐ Đã xác định winsor DV 1st–99th.
☐ Đã xác định winsor firm controls 1st–99th.
☐ Đã xác định macro variables không winsorize.
☐ Đã xác định UCT không winsorize.
☐ Đã xác định Book Leverage formula.
☐ Đã xác định ST Leverage formula.
☐ Đã xác định LT Leverage formula.
☐ Đã xác định Debt Maturity formula.
☐ Đã xác định Market Leverage formula.
☐ Đã xác định treatment của zero-debt firms trong Debt Maturity.
☐ Đã xác định Tangibility.
☐ Đã xác định Profitability.
☐ Đã xác định Size.
☐ Đã xác định Market-to-Book.
☐ Đã xác định GDP Growth.
☐ Đã xác định Inflation.
☐ Đã viết baseline equation.
☐ Đã xác định UCT_t → CapitalStructure_t+1.
☐ Đã xác định Firm FE.
☐ Đã xác định khi nào Year FE được phép.
☐ Đã xác định khi nào Industry × Year FE được phép.
☐ Đã xác định khi nào Country × Year FE được phép.
☐ Đã xác định VCE theo Design.
☐ Đã báo number of year clusters.
☐ Đã báo number of country clusters.
☐ Đã kiểm tra few-cluster issue.
☐ Đã đọc Bui et al. (2026).
☐ Đã hoàn thành Sample Selection Table.
☐ Đã hoàn thành Country Selection Table.
☐ Đã hoàn thành Capital Structure Variable Table.
☐ Đã hoàn thành Identification Diagnostic Table.
☐ Đã hoàn thành Method Decision Matrix.
☐ Đã hoàn thành Model Reference Matrix.
☐ Đã có ít nhất 18–25 Methodology references.
☐ Đã viết Draft Data and Methodology 1.800–2.500 từ.
☐ Mọi citation đã được kiểm chứng.

---

PHẦN AD. THÀNH PHẨM CUỐI CÙNG CỦA BƯỚC 1

Bước 1 chỉ được coi là hoàn thành khi nhóm có đủ:

1. Sample Selection Table
2. Country Selection Table
3. UCT Data Note
4. Capital Structure Variable Definition Table
5. Identification Diagnostic Table
6. Method Decision Matrix
7. Model Reference Matrix
8. Reference List cho Methodology
9. Draft Section 3. Data and Methodology, 1.800–2.500 từ

Khi các sản phẩm trên hoàn thành, nhóm chuyển sang:

BƯỚC 2. KIỂM TRA VÀ DIỄN GIẢI KẾT QUẢ MÔ HÌNH CƠ SỞ