A note on dealing secondary data related to Finance and Accounting dissertation projects: for the HK F&A students

A note on dealing secondary data related to Finance and Accounting dissertation projects: for the HK F&A students (information from perplexity.ai)

My question 1: "For doing Accounting and Finance dissertation projects in Hong Kong, one of the main secondary data category to study is those related to the financial report data and the related economic and financial market statistics. Highlight the main sources of this data category as well as the data quality (including data reliability and comparability) involved. Also suggest some good data gathering and data practices on this data category in the context of Hong Kong."

The reply: "Hong Kong accounting and finance dissertations typically rely on three main types of secondary quantitative data: (1) listed firms’ financial reports, (2) macroeconomic and sector statistics, and (3) financial market indicators from regulators and exchanges. For a dissertation, these should be drawn primarily from official or well‑established commercial databases, with clear documentation of how data are selected, cleaned, and made comparable over time and across firms.

​

Main data sources in Hong Kong

• Listed company financial reports

  • HKEXnews (issuer filings: annual/interim reports, announcements, circulars) for all Hong Kong–listed companies, including audited financial statements prepared under HKFRS/IFRS.
    ​

  • HKEX Annual Report Explorer and related guidance, which standardise and digitise disclosure structures and highlight common reporting practices and deficiencies, useful for understanding items and checking consistency.
    ​

• Macroeconomic and sector statistics

  • Census and Statistics Department (C&SD) for GDP, inflation, labour, industry breakdowns, price indices, business tendency surveys and other official economic series.
    ​

  • Hong Kong Monetary Authority (HKMA) “Economic & Financial Data for Hong Kong” and Monthly Statistical Bulletin for interest rates, monetary aggregates, banking statistics, FX, and external positions, aligned with IMF dissemination standards.
    ​

• Financial market and regulatory data

  • HKEX market statistics for trading volume, turnover, index composition, derivatives statistics, and listed product information.
    ​

  • Securities and Futures Commission (SFC) statistics for securities market indicators, licensed intermediaries, fund management business size, and regulatory metrics.
    ​

• International and comparative sources

  • IMF Dissemination Standards Bulletin Board (DSBB) entries for Hong Kong, which document the methodology, timeliness, and revision policies for major Hong Kong macro‑financial series.
    ​

  • Other multilateral or commercial databases (e.g. for cross‑country comparisons) that often repackage data originally produced by HKMA, C&SD, and HKEX.
    ​

Data quality, reliability, and comparability

• Reliability and credibility

  • HKEX‑listed firms’ annual reports are audited and must comply with Hong Kong Financial Reporting Standards (HKFRS), which are largely converged with IFRS, enhancing reliability for empirical work.
    ​

  • HKMA and C&SD statistics follow international standards and are tied to the IMF’s DSBB framework, with explicit documentation of sources, compilation methods, and revision policies.
    ​

• Comparability across firms and over time

  • Use of HKFRS/IFRS supports cross‑firm comparability, but differences remain due to choices around accounting policies, estimates, and non‑HKFRS performance measures (e.g. EBITDA definitions), which HKEX and its guidance explicitly flag as areas where comparability may be limited.
    ​

  • Time‑series comparability can be affected by new standards (for example, HKFRS 18 on presentation and disclosure effective from 2027) and periodic changes in disclosure rules or listing requirements; these require careful attention to restatements and changes in line items.
    ​

• Data revisions and timeliness

  • C&SD publishes “advance estimates” for key aggregates such as real GDP, which are then revised when more complete data arrive; using a consistent vintage (e.g. latest‑available) is important for empirical robustness.
    ​

  • HKMA data in the Monthly Statistical Bulletin and “Economic & Financial Data for Hong Kong” are updated frequently, with clear indication of last revision dates and whether data are seasonally adjusted, which is important when aligning with firm‑level dates.
    ​

• Limitations and measurement issues

  • Market statistics from HKEX and SFC may treat segments differently (e.g. Main Board vs GEM, derivatives vs cash market) and can change definitions of categories over time, which can introduce structural breaks.
    ​

  • Some variables of interest to corporate finance research (e.g. ownership structure, ESG metrics, or segment‑level data) may be disclosed qualitatively or in non‑standard formats, requiring manual extraction and judgement, which can reduce mechanical comparability despite formal HKFRS compliance.
    ​

Good data gathering practices (Hong Kong context)

• Define population and sample very clearly

  • Specify listing board (Main Board vs GEM), sector classification, and time window, using HKEX classifications and listing histories to ensure consistency when firms change board or ticker.
    ​

  • Document inclusion/exclusion rules for firms with very short listing histories, suspended trading, or extreme corporate events (backdoor listings, large mergers), as Hong Kong has frequent corporate restructuring activity.
    ​

• Standardise variables and coding

  • Extract financial statement items from HKEXnews filings according to consistent HKFRS line items, and create derived ratios (leverage, profitability, etc.) using clearly defined formulas that you apply uniformly across firms.
    ​

  • Where companies report non‑HKFRS measures (e.g. “adjusted profit”), always record the underlying reconciliation and, where possible, base core empirical variables on HKFRS numbers to enhance comparability.
    ​

• Align firm‑level and macro/market data carefully

  • Match firm fiscal year‑end data to the closest relevant macroeconomic and financial indicators from HKMA and C&SD (e.g. using annual averages of interest rates or GDP growth over the firm’s fiscal year).
    ​

  • When using high‑frequency market variables (e.g. daily turnover, volatility, or interbank rates), clearly state whether you aggregate to monthly/annual frequency and how you handle holidays and missing days in the Hong Kong market calendar.
    ​

• Manage revisions and restatements

  • Track whether annual reports contain restated comparative figures due to changes in accounting policies or error corrections, and record both original and restated values where relevant for your research design.
    ​

  • For macro and financial series that are revised (e.g. GDP, some monetary aggregates), decide whether you use real‑time (initial) figures or latest‑available, and apply that rule consistently across the dataset.
    ​

• Ensure transparency and replicability

  • Keep a data dictionary describing each variable’s definition, exact source (HKEXnews URL or HKMA/C&SD table code), units, transformations (e.g. log, winsorisation), and any assumptions or manual judgements used in coding.
    ​

  • Store raw downloads separately from cleaned datasets, and maintain scripts or documented procedures for cleaning and merging so that another researcher could reproduce your sample from the original HKEX, HKMA, C&SD, SFC, or IMF files.
    ​

Good data handling practices for quality

• Quality checks and triangulation

  • Reconcile key totals (e.g. total assets, equity, profit) from hand‑collected data with summary figures in companies’ own tables and with high‑level aggregates reported in HKEX or sector statistics where possible.
    ​

  • Cross‑check macro indicators (e.g. GDP growth, inflation) from C&SD with HKMA’s “Economic & Financial Data for Hong Kong” and IMF DSBB descriptions to ensure consistent units, base years, and seasonal adjustment choices.
    ​

• Handling outliers and structural breaks

  • Identify extreme values that may arise from one‑off events common in Hong Kong’s markets (e.g. equity placings, special dividends, listing‑related expenses) using notes to the accounts and HKEX announcements, and decide in advance how to treat these for empirical tests.
    ​

  • Be explicit about structural changes such as adoption of major new HKFRS standards (e.g. HKFRS 16 and upcoming HKFRS 18) and major regulatory or market reforms, and consider including dummy variables or sub‑period analysis to manage breaks in comparability.
    ​

These practices, combined with disciplined documentation, will typically satisfy dissertation expectations for data reliability, comparability, and research transparency in Hong Kong accounting and finance projects."

My question 2: "Regarding our discussion here, suggest what can be the main challenges and difficulties of gathering and analyzing the secondary data to study the following Finance and Accounting research themes in the Hong Kong context in years 2025-2026:

1. IPO

2. Capital structure of listed companies

3. Dividend policy of listed companies

4. Share buyback of listed companies

5. Working capital management of listed companies

6. Corporate governance-related topics of listed companies"

The reply: "For these six Hong Kong themes in 2025–2026, the main difficulties come from (a) regulatory and disclosure changes, (b) fragmented or non‑standard information, and (c) event‑related timing/interpretation problems for empirical testing. Each topic faces some Hong Kong–specific complications because of concentrated PRC issuer exposure, frequent rule consultations, and the post‑pandemic rebound in activity.

​

IPO research

• Data identification and boundaries

  • Distinguishing between different listing routes (Main Board vs GEM, specialist technology listings, secondary and dual‑primary listings) can complicate sample construction and comparability across IPOs and over time, especially as HKEX revises regimes to attract tech and cross‑border issuers.
    ​

  • Many issuers are PRC‑related and also subject to mainland, US, or other overseas listing rules, so aligning prospectus disclosure, accounting standards, and risk factors across jurisdictions is difficult.

• Time‑varying rules and market conditions

  • Ongoing reforms to IPO pricing, allocation, and free‑float rules affect offer mechanisms and aftermarket liquidity, creating structural breaks in variables like underpricing, subscription ratios, and trading activity before and after 2025 rule changes.
    ​

  • Cyclical swings in IPO volume during the post‑pandemic rebound can lead to clustering by year and sector (e.g. tech, AI), making it hard to disentangle regulatory effects from changing issuer mix and macro conditions.
    ​

Capital structure of listed companies

• Measurement and classification issues

  • Differences in how firms classify hybrid instruments, perpetuals, convertible bonds, and shareholder loans under HKFRS affect leverage measures and the split between debt and equity, reducing comparability across firms and years.

  • Frequent use of offshore vehicles, cross‑border financing and related‑party funding (e.g. from PRC parents) makes it hard to capture the true economic leverage of Hong Kong–listed groups from consolidated statements alone.

• Dynamic environment and endogeneity

  • Listed firms’ capital structures are influenced by evolving mainland–Hong Kong capital‑flow regimes, interest‑rate cycles and sector‑specific policies, which change between 2025 and 2026 and complicate attribution of causality.

  • Firms often undertake multiple overlapping transactions (rights issues, placings, bond issues, share buybacks) in a short period, so isolating the impact of any single capital-structure decision using secondary data is challenging.

Dividend policy of listed companies

• Event identification and consistency

  • Dividend information is dispersed across results announcements, annual reports, and corporate actions; differing fiscal year‑ends and announcement dates complicate the construction of clean dividend event panels.

  • Special, interim, and scrip dividends are sometimes presented in narrative form or combined with other distributions, so misclassification can occur when coding payout levels and stability indicators.

• Structural and sample issues

  • Payout decisions often respond to cross‑border regulatory developments, capital‑control considerations and group‑level cash needs (especially for PRC‑headed groups), which are not directly observable in public data.

  • Survivorship bias arises if delisted firms and very small issuers with irregular dividends are dropped, potentially overstating the stability and level of dividends among Hong Kong‑listed firms.

Share buyback of listed companies

• Data extraction and granularity

  • Buyback actions are typically reported via daily or periodic announcements rather than in easily downloadable structured tables, so constructing a complete and accurate buyback database requires intensive manual collection and cleaning.

  • Distinguishing between authority approvals, actual repurchase executions, and cancellations is non‑trivial; secondary data may not clearly separate intended versus completed buybacks.

• Interpretation and confounding events

  • Buybacks often coincide with other corporate actions (e.g. option exercises, placements, or insider trading windows), making it hard to attribute price and liquidity effects to buybacks alone.

  • Regulatory changes to public float requirements and capital management flexibility around 2025 alter the motivation and constraints for buybacks over time, introducing policy‑driven breaks in behaviour.
    ​

Working capital management

• Limited disclosure detail

  • Financial statements provide only aggregate line items (inventories, trade receivables/payables) by reporting date, so high‑frequency variations in working capital and intra‑year policies are unobservable using standard secondary data.

  • Segment‑level or geographic breakdowns of working capital are often missing or unevenly disclosed, problematic for diversified groups with substantial mainland operations.

• Industry heterogeneity and external shocks

  • Hong Kong hosts a mix of asset‑light service firms and asset‑heavy trading/manufacturing intermediaries; standard working‑capital ratios are not directly comparable across these business models without extensive industry‑specific adjustments.

  • Supply‑chain disruptions, changes in PRC trade policies and interest‑rate movements can all affect working‑capital metrics, but publicly available data rarely allow clean identification of these external drivers.

Corporate governance topics

• Qualitative, non‑standardised data

  • Many governance variables (board processes, risk oversight, ESG integration, internal controls quality) are described qualitatively in Corporate Governance Reports and ESG reports, which vary in length, style, and specificity across issuers.

  • Coding such narratives into consistent quantitative indicators is labour‑intensive, requires judgement, and raises concerns about reliability and inter‑coder consistency.

• Evolving rules and disclosure expectations

  • HKEX and the SFC have been updating governance and ESG‑related requirements, leading to step‑changes in disclosure volume and content around consultation and implementation dates, which introduces breaks in time‑series analyses.
    ​

  • Compliance‑driven “box‑ticking” disclosures can inflate apparent governance quality without necessarily reflecting substantive change, so measured indices based on secondary disclosures may weakly capture real governance practices.

These challenges do not make the topics unsuitable, but they do mean a dissertation should explicitly discuss data limitations, carefully document coding decisions, and, where possible, supplement quantitative secondary data with clearly defined qualitative or case‑study components."

A collection of blog notes on using chatgpt for research purpose.