This project extends the methodology of The Structure of London Financial News to the Russian financial press of the imperial period. Where the London project analyzes the Financial Times and Investor's Monthly Manual (1869–2008), this companion project examines Vestnik Finansov, Promyshlennosti i Torgovli (Herald of Finance, Industry, and Trade) — the official weekly journal of the Russian Ministry of Finance, published in St. Petersburg from 1883 to 1917.
Vestnik Finansov is one of the basic primary sources for Russian economic history of the late imperial period. Each issue contained four sections: statistical tables (exchange rates, bank balances, commodity prices), analytical articles by prominent economists such as I. M. Kulisher and A. A. Isaev, a chronicle of events in finance, banking, agriculture, industry, and trade, and official announcements including company reports for publicly disclosed enterprises.
We have digitized and OCR'd 77,510 pages containing 62 million words from Vestnik Finansov, covering 30 years from 1869 to 1916. The corpus is fully searchable in Russian and English via a custom full-text search interface with FTS5 indexing, bilingual query translation, and frequency visualization. The next phase applies the NLP pipeline developed for the London corpus: NMF topic modeling on TF-IDF matrices, FastText subword embeddings for handling degraded OCR, and Procrustes-aligned diachronic embeddings to track semantic change in Russian financial vocabulary from the 1880s through the Revolution.
This work complements the St. Petersburg Stock Exchange Project at Yale's International Center for Finance, which provides monthly price data for all securities listed on the SPSE from 1865 to 1914. Together, the price data and news corpus enable a comprehensive analysis of information and markets in imperial Russia.
Our approach follows Bybee, Kelly, Manela, and Xiu (2024), "Business News and Business Cycles," Journal of Finance 79, 3105–3147, as adapted for the London Financial News project.
Primary Corpus: Vestnik Finansov, Promyshlennosti i Torgovli (1869–1916). 77,510 OCR'd pages totaling 62 million words across 30 years, sourced from HathiTrust (Princeton holdings) and partner collections. Full-text searchable via SQLite FTS5 with bilingual English–Russian query support.
Supplementary Corpus: Birzhevye Vedomosti (Stock Exchange News, 1862–1917), a St. Petersburg daily with commercial orientation. Partial digitization available via East View / CRL Imperial Russian Newspapers (825,000+ pages, 35 titles, open access).
Topic Model: NMF on TF-IDF, adapted for Russian-language OCR with Cyrillic-aware FastText subword embeddings (character 3–6 n-grams). Russian morphological preprocessing via pymorphy2.
Semantic Change: Per-decade FastText models aligned via orthogonal Procrustes rotation, tracking drift in financial vocabulary (aktsiya, obligatsiya, kredit, birzha) across the reform, industrialization, and revolutionary periods.
Companion Data: SPSE monthly prices (1865–1914), enabling joint analysis of news content and market returns — paralleling the FT/IMM analysis of London markets.
The digitization draws on multiple sources:
The project encompasses two complementary Russian-language financial publications from the imperial period.
| Published | 1883–1917 |
| Corpus Coverage | 1869, 1885–1896, 1898, 1900–1916 |
| Frequency | Weekly (Sundays), 52 issues/year |
| Publisher | Russian Ministry of Finance |
| Location | St. Petersburg / Petrograd |
| Language | Russian (pre-reform orthography) |
| Corpus Size | 77,510 pages · 62,267,974 words |
| Sections | Statistics, Articles, Chronicle, Announcements |
| Status | OCR complete · Full-text searchable |
| Published | 1862–1917 |
| Frequency | Daily (from 1885) |
| Publisher | Stanislav Propper |
| Location | St. Petersburg |
| Language | Russian |
| Content | Commercial news, market reports, financial commentary |
| Status | Partial open access via East View/CRL |
The period 1883–1917 spans the great Russian industrialization under Finance Ministers Bunge, Vyshnegradsky, and Witte; the adoption of the gold standard (1897); the banking crisis of 1899–1901; the Russo-Japanese War (1904–05) and Revolution of 1905; the Stolypin reforms; and the upheavals of World War I through the February Revolution. Vestnik Finansov was the official organ through which the Ministry of Finance communicated policy, published statistics, and shaped the financial narrative of the empire.
Select topics to plot their monthly attention over time. Each series shows the average NMF topic weight across all issue-documents published that month. 25 topics extracted via Non-negative Matrix Factorization on TF-IDF (10,000 features).
Topics clustered by the co-movement of their monthly attention, after removing secular trends. We compute first differences of each topic's monthly attention series, then measure pairwise correlations to capture which topics move together in response to events — financial crises, policy changes, wars, and political upheaval. Hierarchy via Ward's agglomerative clustering.
Pairwise correlations of first-differenced monthly topic attention. Red = topics whose coverage increases and decreases together. Blue = topics that substitute for each other.
How has the meaning of Russian financial vocabulary changed from the 1880s through the Revolution? We train separate FastText embedding models for each decade (100-dim, skip-gram, char 3–6 n-grams) and align them via orthogonal Procrustes rotation, then measure semantic drift as cosine distance. Shared vocabulary: ~24,000 words across 4 decades.
Words ranked by cosine distance between their 1880s and 1910s embeddings. Labels show the decade of the single largest shift.
Select a word to see how its nearest neighbors change across decades. When a word's neighbors shift, its meaning has shifted.
How does the official Russian financial press write, as distinct from what it writes about? Following the London project's journalism style analysis, we measure the rhetorical posture of Vestnik Finansov: the balance of statistical reporting vs. analytical narrative, official announcements vs. critical commentary, and the evolving role of the state-directed press in shaping financial information during Russia's rapid industrialization.
The London project found a 120-year shift from terse market records to narrative-driven analysis — attribution and metaphor rose dramatically. Russia's Vestnik Finansov, as an official government organ rather than a commercial newspaper, may show a fundamentally different trajectory: state-directed information control vs. market-driven journalism.
Six dimensions of journalistic practice measured via curated Russian word lists (pre-reform orthography). All series smoothed over 12 months.
Conflict and metaphor in Russian financial journalism — how does the state-directed press handle crisis and market upheaval?
Numeric tokens per 1,000 words — does the journal become more or less data-intensive over time?
| Dataset | Description | Link |
|---|---|---|
| SPSE Monthly Prices | Monthly price data for all securities, St. Petersburg Stock Exchange (1865–1914) | Yale ICF |
| London Financial News | Topic attention, semantic change, and journalism style from the FT and IMM (1869–2008) | Companion Project |
| Imperial Russian Newspapers | 825,000+ pages from 35 titles including Birzhevye Vedomosti (open access) | East View / CRL |
| Goetzmann & Huang (2018) | "Momentum in Imperial Russia" — JFE paper using SPSE data with 1893 regulatory change | NBER WP |
| Component | Description | Status |
|---|---|---|
| OCR Corpus | Full-text OCR of Vestnik Finansov — 77,510 pages, 62M words (1869–1916) | Complete |
| Full-Text Search | SQLite FTS5 index with bilingual English–Russian query translation (150+ financial terms) | Complete |
| Page Classification | Editorial, statistical tables, securities prices, advertisements, weather | Complete |
| English Translations | AI-powered on-demand page translation via Claude API | Available |
| Research Reports | Speculation in securities (1892–1896), Witte-era finance | Complete |
| Monthly Topic Attention | 25 NMF topics, monthly average weights (1869–1916, 302 months) | Complete |
| Topic Taxonomy | Correlation matrix and Ward clustering of first-differenced topic attention | Complete |
| Semantic Change | Diachronic FastText embeddings by decade (Cyrillic) | Next phase |