I

Despite a wealth of evidence suggesting that well-developed financial markets improve prospects for economic growth (e.g. Levine Reference LEVINE1997), we know relatively little about firm financial strategies in less developed economies, especially in the past. This article investigates corporate finance in such a setting, late Imperial Russia, through the analysis of a novel panel database of annual balance sheet information, governance characteristics and stock prices. Russian corporations faced a number of institutional constraints, such as restricted entry into the corporate form, weak investor protections and thin markets for long-term financing. Yet, despite these obstacles, recent evidence suggests that incorporated firms were particularly important drivers of early industrial modernization in the Russian context (Gregg Reference GREGG2020). We, therefore, address the following question: how did firm characteristics and features of the Imperial legal and financial systems impact corporate financial strategies?

Our unique panel data allow us to link corporate fundamentals and institutional features to their capital structures and (dividend) payout policies. In doing so, we focus on how these firm decisions may have reflected internal agency issues, information asymmetries with external investors, life cycle considerations and other frictions present in the Russian context. Our results demonstrate that Imperial corporations adjusted their financial strategies in the face of these forces and did so in the directions consistent with theory and evidence on firms in more advanced economies. We further show how these factors underlying capital structure and dividend policies played a role in financial performance.

As financial development occurred, leading industrial economies also reduced barriers to forming corporations over the late nineteenth and early twentieth centuries. However, Imperial Russia retained a costly system of incorporation, where each application was potentially subject to intense and heterogeneous scrutiny by Ministry of Finance officials, similar in spirit to entry barriers faced by firms in many developing countries today (e.g. Djankov et al. Reference DJANKOV, LA PORTA, LOPEZ-DE-SILANES and SHLEIFER2002). Owen (Reference OWEN1991) and others have argued that these constraints on forming corporations significantly impeded late Imperial Russian economic growth, as Russian firms could not fully benefit from the legal form's possible financial advantages to adopt modern capital-intensive production technologies.Footnote ¹ Indeed, the recent work of Gregg (Reference GREGG2020) finds a causal relationship between incorporation and firm growth. However, what is missing from that analysis, and from much of the literature on the corporation, is evidence on how this growth was financed. This motivates our exploration of novel balance sheet information to document and analyze Russian corporate finance in this period.

Costly chartering was one factor that limited the number of corporations in Imperial Russia, relative to countries with more liberal incorporation mechanisms.Footnote ² The individualized chartering process, moreover, also resulted in substantial firm-specific differences in activities, governance and managerial characteristics, each of which may have influenced subsequent capital structure and payout decisions. How did these organizational elements interact with the specifics of the Imperial financial system to generate such outcomes? Using a single cross-section from 1914, Gregg and Nafziger (Reference GREGG and NAFZIGER2019) conclude that incorporated firms in Imperial Russia showed considerable adaptability with respect to capital structure (i.e. debt vs. equity) and dividend payout decisions by focusing on variation across three dimensions: governance structure, age and sector.

Taking advantage of our richer balance sheet panel data, merged to stock market information and information on founder characteristics, we first verify that Russian corporations could adjust their capital structures and payout policies according to their needs. Then, this article unpacks financial, agency, signaling and other motivations for using credit and issuing dividends, and we document how these particular dimensions mattered for the financial performance of industrial corporations. Our analysis considers all chartered non-financial corporations in the Russian Empire between 1899 and 1914. We compile the panel of financial data from yearbooks of the Ministry of Finance and match these to the characteristics of corporate founders and basic governance indicators from the RUSCORP database of corporate charter information (Owen Reference OWEN1992), and with the final monthly share prices for listed firms on the St. Petersburg stock exchange (Yale ICF). Relative to our earlier article that examined a single-year cross-section of fewer than 1,500 observations, this much larger panel dataset of almost 20,000 observations over more than 14 years enables more rigorous evaluations of different corporate finance theories in the Russian context, improves statistical inference and offers a sharper picture of the development of the corporate sector during a key period of economic change.Footnote ³ Thus, these data present a unique opportunity to explore the linkages between corporate structure and financial behavior among early industrial corporations.

Though Imperial Russia had a substantial banking system and active securities exchanges and was well integrated with European capital markets, Russian corporations still faced significant information asymmetries, agency issues, and missing markets. In this context, we show that corporations that were closely held, that were younger, that listed on stock exchanges and that were larger in terms of total assets all tended to rely more on debt financing. Moreover, corporations with more tangible assets tended to have less debt, suggesting that tangible assets did not help Russian corporations access less expensive loans. These findings suggest the kinds of external conditions and internal fundamentals that motivated Russian corporations to choose particular capital structures. In particular, the Russian stock markets functioned to allow those corporations that listed to reduce their leverage, credit institutions helped to finance short-term expenses like inventories but not fixed property, and asymmetric information within the firm impacted capital structure in ways similar to what is emphasized in the modern corporate finance literature (i.e. Harris and Raviv Reference HARRIS and RAVIV1991; Graham Reference GRAHAM2022). Corporate dividend issuance similarly reflected tradeoffs emphasized in the broader literature: corporations with higher levels of debt had lower payout ratios, corporations at later stages of the life cycle tended to pay out a greater proportion of profits and more closely held corporations used dividends as a means of returning profits to owners. Furthermore, we show, using a dynamic model of dividend smoothing, that corporations adjusted how they issued dividends in response to changes in earnings.

We conclude by showing that variation in a firm's ownership structure, age and interactions with securities markets mattered for two measures of financial performance: firms’ return on assets and market-to-book ratios. Corporations that listed shares on the St. Petersburg stock exchange had higher returns on assets, which reflected positive selection into listing or the need for additional equity in making productivity-enhancing investments. Firms with a more closely held corporate form had higher average returns on assets but smaller market-to-book ratios. Moreover, corporations that issued higher dividends saw greater subsequent returns on their assets, which suggests that dividends embodied a signal about future performance. Furthermore, higher dividends as a share of profits were also associated with a higher market-to-book ratio. We interpret such findings to mean that corporations with more tightly controlled structures, likely facing fewer costly principle–agent governance problems, enjoyed higher returns on assets (potentially paid out via dividends). Moreover, higher market returns compensated for agency issues in widely held corporations, and dividends compensated for poor investor protections.

In this article, we first outline the relevant institutional, economic and financial characteristics of the late Imperial economy and the nascent corporate sector. This provides a starting point for thinking about the underlying drivers of Russian corporate capital structures and payout policies as suggested by the context and modern corporate finance theory. We then present our new database and document patterns in balance sheet characteristics across different types of corporations and over time. The empirical work that follows investigates the determinants of corporate leverage and dividend payout strategies. We close by considering how corporate governance, life cycles and other factors influenced financial performance. Our conclusion offers some broader takeaways for the financing of early industrialization and suggestions for future research.

II

We focus on the Russian economy between the late 1890s and World War I. According to the national income and business cycle research of Gregory (Reference GREGORY1982) and Owen (Reference OWEN2013), over this period the Russian economy experienced a mid-decade boom, followed by a downturn (bottoming out in 1901), growth to 1905, a massive contraction with the 1905 Revolution, and a slow, erratic recovery leading up to World War I (see Figure 1, Panel A). While per capita income changed little over the period and the economy remained largely agrarian, this period did see substantial early growth of Russian industry (Gregory Reference GREGORY1982). A long line of scholarship interprets this early industrial development as a consequence of various state initiatives in the economy, supported by growing foreign investment (e.g. Von Laue Reference VON LAUE1965).Footnote ⁴

At the same time, Cheremukhin et al. (Reference CHEREMUKHIN, GOLOSOV, GURIEV and TSYVINSKY2017) assert that late Imperial industrialization was constrained by excessive market power. This is consistent with the research of Owen (Reference OWEN1991) and Gregg (Reference GREGG2020), who argue that the absence of general incorporation constrained firm expansion and output growth in this period. In this interpretation, corporations possessed numerous advantages relative to other firms, particularly when it came to addressing agency issues and accessing cheaper sources of external funding. Indeed, incorporated firms were at the heart of the modernizing subsectors of Russian industry, and they were prominent participants in the nascent Russian financial system (Kulikov and Kragh Reference KULIKOV and KRAGH2019; Shepelev Reference SHEPELEV1973). However, we know very little – especially quantitatively – about how Imperial industrial corporate finance worked in practice. To contextualize and motivate our effort to do just that, we outline key features of the historical environment and connect these to insights drawn from modern corporate finance theory.

The Imperial Russian financial system

The financial environment in late Imperial Russia structured the options faced by corporations. In practice, Russian companies could finance operations or expansion through retained earnings, informal sources of credit and equity investment, formal loans from a nascent banking sector, or access to thin but growing bond and stock markets. For particularly large and successful firms, limitations of domestic sources of financing led them to turn to Western European banks and securities markets. While the options available to Russian corporations resembled those in other industrializing economies of the time, context-specific conditions likely impacted the relative and absolute costs of financing through different sources, with implications for financial strategies.

In general, Imperial Russia is commonly viewed as possessing weak financial markets and institutions. According to Rajan and Zingales (Reference RAJAN and ZINGALES2003), Russia had a low bank deposits-to-GDP ratio, few exchange-listed firms, and low stock market capitalization-to-GDP ratio in 1913.Footnote ⁷ Alexander Gerschenkron famously doubted Russian banks’ ability to provide meaningful financial assistance to industrial enterprises (Reference GERSCHENKRON1962, pp. 19-20). Yet, when we consider the Imperial financial system in a broader comparative perspective (Table 1), we find that Russia occupied a position similar to other European countries and ahead of many peripheral ones. For example, Russia's ratio of financial assets to GDP was greater than or similar to that in the United States, Canada, the United Kingdom and France, larger than in Spain, Argentina and Brazil, and significantly behind only Germany and Sweden.Footnote ⁸ Despite constraints of the concession system, corporations’ high levels of capitalization and outsized role in the economy meant that Russia's ratio of equity issues to total capital formation in 1913 was the second highest of the countries in Table 1. Finally, though its deposits-to-GDP ratio was low in 1913, Russia's ratio of total loans to GDP is among the highest in Table 1, only falling behind what prevailed in Germany and France.Footnote ⁹ Thus, Russian banks and securities markets apparently generated a significant supply of financing by the early twentieth century.

Table 1. Comparative indicators of financial development c. 1913

^a Rajan and Zingales (Reference RAJAN and ZINGALES2003). Deposits are commercial + savings deposits.

^b Fohlin (Reference FOHLIN2012, tables 6.1, 6.2 and 8.2); for universal banking and equity markets, this table reproduces the subjective ranking of 0, 1, or 2 (0 – least; 2 – most; latter) from that source. ‘Financial institution assets’ include those of banks, other credit institutions, insurance companies, savings societies (including pensions), insurance companies and others.

^c Goldsmith (Reference GOLDSMITH1969a) as reported in Musacchio (Reference MUSACCHIO2009, p. 66), or derived directly from the former (UK, Russia); ‘loans’ are from all financial institutions and include mortgages

^# - The first number includes the Polish contribution to the Russian Empire's GDP in the denominator; the second does not. The Russian entry for this variable is derived directly from Goldsmith (Reference GOLDSMITH1969b) and Gregory (Reference GREGORY1982).

^A Great Britain; ^B England

Recent research suggests that Gerschenkron may have underestimated the efficacy of the Russian banking and financial systems. Investment banks provided industrial firms with payment and discounting services, and with special drawing accounts (onkoli) collateralized by various securities (see below).Footnote ¹⁰ In contrast, longer-term bank credit played a relatively minor role in firm finance, although the rolling over of short-term financing was prevalent. The State Bank and affiliated entities took deposits, issued some loans, and discounted bills of exchange for industrial firms through provincial branches and local treasury offices. Municipal banks, credit societies and other savings institutions played a more limited role in industrial finance, although they did hold corporate securities on their balance sheets. However, despite these varied sources of credit available to Russian firms, credit in the Russian Empire remained expensive relative to other nearby European markets. For example, in 1900, the market discount rate in St. Petersburg was 6.68 percent, where the open market discount rate that year was about 4.41 percent in Berlin and 3.04 percent in France (Homer and Sylla Reference HOMER and SYLLA2005, pp. 228, 265 and 605).

Imperial Russian securities markets were dominated by government and land-related securities, including state-backed railroad debt, notes issued by land banks and the mortgage-like bonds that financed serf emancipation. However, commercial banks facilitated the placement of corporate debt and equity, often executing this by holding such assets on their books and issuing ‘shares’ in these accounts to the public (Crisp Reference CRISP1976, pp. 144-6). This form of intermediation likely eased the costs of information asymmetries between firms and investors, especially for firms with less tangible (and therefore collateralizable) assets. The period from 1861 to 1914 saw steady growth in the number of formally listed securities and the total market capitalization of firms whose shares were traded on domestic exchanges.Footnote ¹¹ Supporting these developments, from the mid 1890s, state banking institutions increased deposits in joint-stock banks, thus fostering an implicit guarantee for many of the securities held on the books of the banking system.Footnote ¹² In combination with the concession system of incorporation and the possible role of government procurement policies, this is suggestive that the political and social ties of corporations could impact their financing options, which we explore with our data.Footnote ¹³

Thus, Imperial Russian non-financial corporations could raise funds for expansion or operations through ‘family and friends’, retained profits, direct credit (often in the form of short-term drawing accounts), the issuance of debt securities, or the selling of new equity. Given the apparent prevalence of foreign capital in these channels (e.g. Crisp Reference CRISP1976), substantial parts of the Russian financial system accessible by the corporate sector were linked to intermediaries and securities’ markets in Western Europe. Large geographic distances, high transportation costs and relatively thin trading potentially meant that external informational constraints were significant for corporations facing both foreign and domestic potential investors. Such information asymmetries were likely only partially mitigated by the financial press and the monitoring (and underwriting) of financial intermediaries. Moreover, these frictions were compounded by the general lack of de facto legal protection for small and medium investors, even after a 1901 reform explicitly addressed this issue (Gregg Reference GREGG2017).Footnote ¹⁴

Unfortunately, we have little direct evidence on how expensive the different sources of available financing really were (or, alternatively, how financially constrained firms were in practice). However, our panel balance sheet data make it possible to document how capital structure, payout policies and financial performance varied over time (and over the business cycle), across industries and among corporations of different ages, sizes, asset compositions, governance structures, political connections and profitability. These empirical relationships can be plausibly linked to underlying agency issues, investment opportunities, external information asymmetries, transaction costs, or financial supply-side conditions faced by corporations.

III

How can we interpret evidence on the financial decisions of Imperial Russian corporations? This section highlights theoretical and empirical work in modern corporate finance that may help explain how firm fundamentals and features of the Russian historical context mattered for capital structure and dividend issuance. Variations in these two attributes are important indicators of the ways that firms address internal and external constraints on financing their operations and growth, with potential implications for corporate valuation and performance.

IV

This article draws on a panel of newly compiled balance sheet data on all Imperial Russian non-financial corporations active from 1899 onwards.Footnote ³⁰ We collected data for individual corporations as reported in the Ministry of Finance Yearbooks published from 1900 through 1915. Then, we matched companies over time by hand to form the panel. We also matched companies by corporation name to the RUSCORP database (Owen Reference OWEN1992) to exploit the information on initial chartered characteristics of the corporations in that source, such as the corporation's type (A-corporation vs. share partnership, as denoted by the use of different terms for ‘share’) and the location (region) of the headquarters. RUSCORP also provides data on the personal characteristics of all corporations’ founders, as listed in the charters, which can be used to define whether a corporation has a founder who is a government official, noble, or member of the gentry.Footnote ³¹ Finally, we match by corporation name to the monthly security prices on the St. Petersburg stock exchange.Footnote ³² From these observations, we calculate within-year share price volatility and average yearly share prices, and estimate the annual corporate valuation as that price times the number of shares at founding.Footnote ³³ While this may introduce some measurement error, as corporations could have changed their numbers of shares after founding, unfortunately, we have found no source listing both a company's market share price and its current number of shares.

As we noted above, the Ministry of Finance compiled the balance sheet information in their yearbooks from the official commercial periodical Vestnik finansov i torgovli, in which corporations issued financial statements required by the commercial code and by their individual charters. Figure A1 in the online Appendix presents entries for the Martens and Daab Partnership in the 1901 accounting year, which show that the publicly announced information matches what we find in a codified form in the 1902 yearbook. Other such spot checks suggest that the yearbooks did accurately consolidate data from the Vestnik periodical, although we have no way to check the underlying quality of the publicly issued balance sheets in the latter source.Footnote ³⁴

We construct our panel dataset from balance sheet information for the accounting years 1899–1914, with some observations from earlier years. We extract variables related to assets, liabilities and cash flows (including profitability); details and summary statistics are provided in the online Appendix (Tables A2–A5). In its entirety, the resulting dataset describes 2,874 unique corporations observed in at least one year for 19,817 observations. This represents almost 70 percent of the total non-financial corporations established in Imperial Russia.Footnote ³⁵ The implied annual number of corporations in our database was relatively stable (1,100–1,700) except for some reporting of earlier accounting years in the 1900 Ministry of Finance yearbook and the low numbers for the disrupted year of 1905.Footnote ³⁶ Textiles, foods and metals represent the largest industrial categories, reflecting required capital intensity and the large size of the food sector. Mining, which was also capital-intensive, is well represented in the database.

Our data include corporations in a variety of sectors with very different capital requirements, market structures and demand patterns. As such, we expect to find substantial differences in financial strategies across industries, scaling by underlying differences in firm size (total assets). Table 2 shows that this is the case. Corporations varied greatly in their property, credit, assets (relative to share capitalization) and profits across industries. The industries with the most property relative to total assets were the municipal services (infrastructure), mining and transportation industries, but those industries did not necessarily have a large amount of credit relative to assets, suggesting a greater reliance on equity and retained earnings. The most profitable industries tended to be newer, more technologically advanced industries of the Second Industrial Revolution, such as chemicals and transportation, though textiles was also quite profitable. These key balance sheet items also changed quite a bit over time, as demonstrated in the figures of Panel B.Footnote ³⁷ Corporate property declined after the 1905 Revolution, while, in general, total assets and credit increased each year. The pre-1910 profits as a share of assets showed a downward trend until 1905, when it began to increase steadily until the end of the period. All of these trends may reflect significant credit expansion and investment in building firm assets over this early period of industrial development. Finally, Panel C indicates the large average financial differences between the two Russian corporation types. Unconditionally, closely held corporations (share partnerships) were much more likely to finance operations out of credit, despite having lower levels of real property. Moreover, even without controlling for industry or age, closely held corporations appear slightly more profitable on average. Given the large differences across industry, year and corporation type demonstrated by Table 2, we turn to examine these dimensions in a multivariate regression framework below.

Table 2. Summary statistics by industry, year, and corporation type: balance sheet items

Source: Ezhegodnik ministerstva finansov [Ministry of Finance Yearbook], 1900-15. In all panels, Profit in 1910 is ‘Balance profit’, and Profit after 1911 is ‘Profits for distribution’. Here, a ‘widely held’ corporation is one that uses the word aktsiia for ‘share’. Corporations with values of 0 for total assets are excluded.

V

In this section, we examine a variety of standard debt ratios to understand the basics of Imperial Russian corporate capital structures. Our analysis is not exhaustive in examining every factor driving capital structure decisions, but we focus on key dimensions that are reflected in our data and suggested by the modern corporate finance literature. Moreover, these exercises are descriptive in nature, as our outcomes and a number of the right-hand-side variables were likely jointly determined by corporations making their capital structure decisions. Following our discussion in Section III, we estimate variants of:

(1)$$\eqalign{y_{it} &= \beta _0 + \beta _1logAssets_{it} + \beta _2\left({\displaystyle{{Profits} \over {Assets}}} \right)_{it} + \beta _3logAge_{it} + \beta _4Listed_{it} \cr &\quad + \beta _5\left({\displaystyle{{Property} \over {Assets}}} \right)_{it} + \beta _6WidelyHeld_{it} + Industry_{ij}^{\prime} {\boldsymbol \gamma } + Region_{ij}^{\prime} {\boldsymbol \delta } + \mu _i + \zeta _t \cr &\quad + \epsilon _{it}}$$

In this regression, y _it is a measure of corporation i's leverage in year t, defined as versions of the credit/asset ratio. In the online Appendix (Table A7, Panel D), we consider the book value of leverage and an estimate of the market value of leverage. Our main right-hand-side variables represent factors suggested in Section III as important for a Russian corporation's capital structure: size (total assets), profitability, age, asset tangibility, whether the corporation lists shares on the St. Petersburg stock exchange, and whether the corporation is widely held (proxied by A-corporation status). We control for industry and region to account for differences in fixed sectoral and geographic components of the demand for and supply of financing. To condition on macroeconomic factors and changes in reporting, we also include accounting year fixed effects (ζ _t). We estimate this regression using random effects and fixed-effect (μ _i) panel specifications, where in the random effects regressions we cluster standard errors by firm ID, and, in the fixed effects regressions, we cluster by industry.Footnote ³⁸

Additionally, we investigate how founder connections and market-to-book ratios relate to corporate capital structure. It may be the case that corporations with members of the government, nobility, or gentry had access to additional sources of financing, whether credit or equity, which impacted capital structure. Furthermore, as a proxy for investment opportunities, we utilize the market-to-book ratio, measured as the firm's total market valuation divided by the par value of share capital, where valuation is the current share price times the corporation's number of shares at founding. We use the initial number of shares in both numerator and denominator because we do not know how many new shares the firm issues after its initial chartering. This variable is defined for those corporations listed on the St. Petersburg exchange.

Table 3, Panel A presents our baseline regressions for Imperial Russian corporate leverage, measured with debt and credit ratios.Footnote ³⁹ We find that whether a corporation was listed on the St. Petersburg exchange and whether it was widely held were both negatively associated with credit and debt ratios; implying that such corporations relied relatively more on equity finance (confirmed in columns 1 and 2 of online Appendix Table A6). On average, listed corporations held almost 18 percent less credit as a share of assets, and widely held corporations held almost 24 percent less credit as a share of total assets (column 1). A corporation's age is negatively related to its use of credit, which is consistent with its role as a proxy for (declining) investment opportunities. For the subset of corporation-year observations for which we can construct the market-to-book ratio, we find little relationship with our leverage measure (column 5), although our sample size is greatly reduced. Size, as measured by the logarithm of total assets, is strongly and positively associated with credit and debt ratios.Footnote ⁴⁰ Larger firms may have engaged in less risky projects, had more collateral on hand, or faced relatively greater short-term financing needs. Across Panel A, profits as a share of assets is negatively related to credit or debt, which echoes pecking order theories of capital structure rather than the tax concerns of tradeoff models. The relationships we find between leverage and size, profitability, and listing also hold in regressions controlling for corporation fixed effects (column 4), though the relationships with age and tangible assets lose statistical significance.Footnote ⁴¹ Finally, column 6 presents estimates that use a 10-year balanced panel, with similar results to the baseline estimates.

Table 3. The underpinnings of imperial Russian corporate debt, credit, and leverage

*** p < 0.01, ** p < 0.05, * p < 0.1. Standard errors clustered by firm ID in parentheses, except in column 4, where standard errors are clustered by industry (fixed from the firm's first observation). The balanced panel in column 6 includes only observations present each year from 1899 to 1909. Column 4 reports the overall R². A corporation is ‘listed’ if its shares appear on the St. Petersburg stock exchange that year. A corporation is ‘widely held’ if it uses the term aktsiia for ‘share’. Founder connections (‘Corporation has noble’, etc.) are coded by matching to RUSCORP (Owen Reference OWEN1992), which includes demographic information for corporate founders. The market-to-book ratio is calculated as the market share price times the number of shares at the corporation's founding divided by total share capital. Log() denotes the natural logarithm.

*** p < 0.01, ** p < 0.05, * p < 0.1. Standard errors clustered by firm in parentheses. A corporation is ‘listed’ if its shares appear on the St. Petersburg stock exchange that year. A corporation is ‘widely held’ if it uses the term aktsiia for ‘share’. Founder connections (‘Corporation has noble’, etc.) are coded by matching to RUSCORP (Owen Reference OWEN1992), which includes demographic information for corporate founders. Log() denotes the natural logarithm.

In Table 3, Panel A, we find that asset tangibility is associated with a lower credit and debt ratio. This may be indicative of the prevalence of short-term borrowing (and the potential importance of maintaining short-run financial flexibility for such firms) or the significant role that agency issues may have played in these corporations.Footnote ⁴² To further explore these hypotheses, we consider several additional definitions of tangible assets in Table 3, Panel B, where column 1 repeats Panel A's column 1 for comparison. For reference, the average value of property as a share of assets was 0.486. Column 1 shows that increasing property/assets by 0.10, or 20 percent of the average value, is associated with a decline in credit/assets of 2.81 percent. In column 2 of Panel B, we use a definition of tangible assets that includes both Property and Goods and Materials (here abbreviated as ‘Inventories’). Now, the relationship between tangible assets and (log) credit/assets is positive in absolute magnitude and statistically significant. Increasing property plus goods and materials as a share of assets by 0.10 (where the average value is 0.671) raises the credit-asset ratio by 4.05 percent on average. In column 3, where our measure of tangible assets is only Goods and Materials (divided by assets), the relationship is unambiguously positive. Increasing inventories by 0.10 (out of an average of 0.185) raises the credit-asset ratio by 12.5 percent on average. Thus, Russian corporate balance sheets show a positive relationship between our scaled measures of credit and inventories, suggesting that the loans likely financed rolling, short-term production expenditures rather than spending on fixed assets like machines and real estate. This is suggestive of the critical importance of maintaining short-run financial flexibility in leverage decisions of Russian corporations in this period, a consideration emphasized by Graham (Reference GRAHAM2022) for modern corporations.

Overall, we find the results in Table 3 (and those in the online Appendix) to be consistent with the capital structure relationships outlined in Section III, particularly those implied by pecking order theories. Asset tangibility was negatively associated with credit, which is consistent with the relevance of agency issues within the corporate sector. This is also supported by the differences in financing by corporation type.Footnote ⁴³ Moreover, across specifications, there is some suggestive but noisy evidence that founder identity influenced access to credit, positively for gentry and negatively for government-affiliated founders. This is not surprising in a setting when personal relationships likely played an important role in corporate founding and in accessing the financial system: perhaps having a government-affiliated founder eased access to equity markets, while a gentry insider could more readily access private credit networks.Footnote ⁴⁴

VI

Internal agency issues, external asymmetric information, profitability and investment are important considerations in corporate payout policies. Was this the case in Imperial Russia? Table A8 in the online Appendix documents how the payout ratio (dividends/profit) varied by industry, over time and by corporation type.Footnote ⁴⁵ Across industries, corporations paid roughly similar dividends as a proportion of profits, though trade and infrastructure corporations paid slightly higher dividends. The dividend/profit ratio varied over time, roughly following the business cycle (see online Appendix Figure A2). Closely held share partnerships tended to pay a greater proportion of profits as dividends than the widely held A-corporations, suggesting that the former rewarded their shareholders directly through dividends, though this type of corporation differed significantly along other dimensions. To better differentiate among the factors driving these bivariate relationships, we turn to multivariate analysis.

We consider the factors associated with corporate dividend/profit ratios in Table 4, where, controlling for industry, the accounting year and the headquarter region, we provide estimates of:

(2)$$PayoutRatio_{it} = \beta _0 + \beta _1\log ( {Assets} ) _{it} + \beta _2\log ( {Creditors/Assets} ) _{it} + \beta _3WidelyHeld_{it} + \mu _i + \zeta _t + \epsilon _{it}$$

Table 4. Factors associated with corporate payout ratios

*** p < 0.01, ** p < 0.05, * p < 0.1.

Standard errors clustered by firm ID in parentheses in columns 1, 2, 4, and 5. Standard errors clustered by industry in parentheses in column 3. Column 3 reports the overall R². Here the payout ratio is defined as the ratio of dividends to profits, trimmed to remove the bottom and top 1% of values. Profits are reported when revenues exceed expenditures. Profit in 1910 is ‘Balance profit’ and Profit after 1911 is ‘Profits for distribution’. A corporation is ‘widely held’ if it uses the term aktsiia for ‘share’. Founder connections (‘Corporation has noble’, etc.) are coded by matching to RUSCORP (Owen Reference OWEN1992), which includes demographic information for corporate founders. Change in dividends is the change in the value of total dividends. Log() denotes the natural logarithm.

Our regressions include plausibly important considerations underlying the variation in Imperial Russian corporate dividend ratios.Footnote ⁴⁶ As we emphasized in Section III, corporate age, size, governance type, leverage and profitability may all have impacted the use of dividends to potentially incentivize managers, signal and/or reward outside investors, and respond to the firm's present financial conditions.Footnote ⁴⁷

The results presented in Table 4 reveal several important sources of variation in Russian corporate dividend policies. The specification in column 1 focuses on three key correlates of dividend issuance: corporation size, credit divided by assets (leverage) and ownership structure, but without any controls for year, location, or industry. Here we see several fundamental relationships: larger corporations, corporations with less debt, and closely held corporations tended to have larger payout ratios. On average, doubling the size of total assets increases the payout ratio by 0.014, a relatively small change. More substantially, transitioning from a closely held to widely held corporation decreases the payout ratio by 0.074 (given an average payout ratio of about 0.415). Larger corporations may have seen more stable earnings and less need to invest out of earnings and hence could pay out more as dividends. The debt finding is consistent with standard debt deductibility stories. The finding on widely held corporations contradicts a simple theory of dividend issuance in this context (as with La Porta et al. Reference LA PORTA, LOPEZ-DE-SILANES, SHLEIFER and VISHNY2000) in which corporations with greater agency concerns would pay higher dividends to reassure shareholders. In this context, where most corporations did not list shares on stock markets, closely held corporations may simply pay higher dividends to the tighter circle of shareholders that both operates and owns the company.

Column 2 includes controls for industry, year and region as well as corporate age, amortization as a share of total assets and indicators for founder connections. We find mixed evidence that age, which could be thought of as a proxy for firm investment demand and firm reputation, impacted corporate payout policy. Similarly, founder connections are not strongly related to payout ratios. However, these estimates show that corporations that devoted more resources to saving for future capital purchases through amortization also tended to have higher payout ratios, since such corporations may pay into amortization when they do not pay down debts (hence decreasing net profits and leading to a higher payout ratio). Similar patterns are apparent in column 4, which includes corporation fixed effects.

Columns 1 through 3 take a static view of dividend payout ratios. To develop a more complete picture of whether corporations’ dividend policies respond to shocks, we consider models of dividend smoothing.Footnote ⁴⁸ Following Fernau and Hirsch (Reference FERNAU and HIRSCH2019) and others, we estimate the following dynamic regression:

(3)$${\rm \Delta }Dividend_{it} = \beta _0 + \beta _1Dividend_{i, t-1} + \beta _2ProfitorLoss_t + \epsilon _{it}$$

Here, −β ₁ is the speed of adjustment (SOA).Footnote ⁴⁹ Leary and Michaely (Reference LEARY and MICHAELY2011) find speeds of adjustment ranging from 0.1 to 0.5 in modern data. We also augment this regression to consider additional key covariates: corporation size (log assets), the log of credit over total assets, ownership structure and the log of amortization over total assets.

In columns 4 and 5, we find speeds of adjustment that sit well within the range found by Leary and Michaely.Footnote ⁵⁰ The additional covariates presented in column 5 suggest that larger, more widely held corporations and corporations that have less leverage and amortization adjust their dividends by larger amounts on average.Footnote ⁵¹ The included covariates, moreover, have economically substantial relationships with the change in dividends, given the average change of approximately 7,000 rubles (column 5). These results show that, rather than adopting some static rule of dividend issuance, Russian corporations adjusted their dividends according to financing conditions, much as documented for modern firms.Footnote ⁵² Taken together, the results for payout ratios lend less support to a model of dividends as signaling mechanisms than to dividend policy as a response to agency concerns and cash flows.

VII

The previous sections have shown that corporations with different profitability, governance structure, access to stock markets and other fundamental characteristics pursued very different financial strategies with respect to capital structure and dividends. We next consider whether differences across firms in these attributes also mattered for their financial performance. While financial outcomes do not directly correspond to productivity or firm growth, reported profitability or market valuations are likely associated with real economic outcomes.Footnote ⁵³

We consider three indicators of financial performance. First, we study the return on assets (ROA), as measured by the ratio of a corporation's profits or losses to its total assets. Second, we consider the return on equity (ROE), measured by the ratio of a corporation's profits or losses to its share capital.Footnote ⁵⁴ Finally, we use the market-to-book ratio, measured as before by the corporation's market share price multiplied by the number of shares at founding, divided by share capital (at par value).

We first estimate the baseline model below, which relates a company's return on assets or equity to its age, a dummy for whether the corporation is listed on the St. Petersburg stock exchange, whether the corporation is widely held, whether the corporation has politically connected founders, and controls for region, industry and year.

(4)$$\eqalign{ROA_{it}\;{\rm or}\;ROE_{it} &= \beta _0 + \beta _1Listed + \beta _2Log( {Age} ) _{it} + \beta _3WidelyHeld_{it} \cr &\quad + \beta _4NobleFounder + \beta _5GovernmentFounder \cr &\quad + \beta _6GentryFounder + Industry_{it}^{\prime} {\boldsymbol \gamma } + Region_{it}^{\prime} {\boldsymbol \delta } + \mu _i + \zeta _t \cr &\quad + \epsilon _{it}}$$

We investigate whether corporations listed on the St. Petersburg stock exchange or those with particular kinds of founders differed in their financial performance to better understand whether differential access to markets or financing mattered for profitability. Given the possibly higher agency costs in the more widely held A-corporations, we might expect such corporations to have lower returns. Firm age captures both selection and differences in the availability of new opportunities over a corporation's life cycle.

Column 1 of Table 5 presents results from estimating our baseline regression (Equation 4) using ROA as the dependent variable. We find a strong negative relationship between the widely held dummy and ROA, suggesting that agency issues may have played a role in lowering returns for corporations with more diffused ownership.Footnote ⁵⁵ Widely held corporations have ROA that is lower than that of closely held corporations by 0.0188 (where the average ROA in the regression is 0.0302). Corporations that were listed on the St. Petersburg stock exchange had higher ROA. The positive relationship between ROA and Listing could simply indicate positive selection into listing, or it could imply that the additional financing provided by trading on the exchange allowed listed corporations to take advantage of high-return opportunities. Older corporations tended to have lower returns on assets, consistent with such corporations facing lower value investment projects. Finally, though corporations with noble or gentry founders do not appear to enjoy any differences in ROA, corporations with government-connected founders do exhibit lower returns. Such government-connected corporations may have faced lower entry barriers in the concession system; hence, such corporations might be observed with lower performance in any cross-section.

Table 5. Performance regressions: corporate return on assets and return on equity

*** p < 0.01, ** p < 0.05, * p < 0.1. Standard errors clustered by firm ID. Return on assets (ROA) is defined as the profit or loss divided by total assets. A corporation is ‘widely held’ if it uses the term aktsiia for ‘share’. Founder connections (‘Corporation has noble’, etc.) are coded by matching to RUSCORP (Owen Reference OWEN1992), which includes demographic information for corporate founders. A corporation is ‘listed’ if its shares appear on the St. Petersburg stock exchange that year. IHS of ROA denotes the inverse hyperbolic sine transformation of our ROA variable. Share price volatility is the annual coefficient of variation of monthly share prices. Log() denotes the natural logarithm. The market-to-book ratio is the calculated as the market share price times the number of shares at the corporation's founding divided by total share capital.

Next, we augment the baseline specification with additional dynamic (lagged) covariates. Columns 2 and 4 show that the level and changes in dividends scaled by total assets tend to predict future profits. In contrast to our findings in Table 4, this suggests that dividends may have conveyed a signal to investors about the corporation's future performance. Column 3 shows that the corporations that were more indebted in the previous period tended to have lower performance in the next period, consistent with the negative relationship between leverage and profitability we found in Table 3. In these columns that include lagged dividends and credit, the relationship between age and ROA changes direction and is even positive and statistically significant in column 4, suggesting that these additional covariates are correlated with age. Overall, the results of columns 2 through 4 suggest that in this context, capital structure and dividend policy were associated with differences in corporate financial performance.

Column 5 presents a baseline result where the dependent variable is ROE. Though the relationships with listing and widely held status are similar to those presented previously, here we see a positive relationship with age. This surprising relationship perhaps arises because older corporations grew in size and thus generated profits that were larger in magnitude; since corporations’ share capital was somewhat fixed, return on equity cannot capture this change as well as return on assets.Footnote ⁵⁶

Finally, we investigate whether these corporate characteristics, along with dividend-profit ratios, were associated with market-to-book ratios. Campbell and Turner (Reference CAMPBELL and TURNER2011) find that nineteenth-century British corporations with higher dividends also had higher market-to-book ratios. They argue that corporations distributed dividends to compensate investors for poor legal protections, thereby increasing demand for equity and raising firm valuations. We speculate that dividends may have served a similar function in the Russian context.

We estimate the following regression model:

(5)$$\eqalign{MB_{it} &= \beta _0 + \beta _1DividendProfitRatio_{it} + \beta _2Age_{it} + \beta _3WidelyHeld_{it} + Industry_{it}^{\prime} {\boldsymbol \gamma }\cr &\quad + Region_{it}^{\prime} {\boldsymbol \delta } + \mu _i + \zeta _t + \epsilon _{it}}$$

We measure the market-to-book ratio as above. Column 6 of Table 5 shows that the dividend/profit ratio was positively related to the market-to-book ratio, suggesting that Campbell and Turner's (Reference CAMPBELL and TURNER2011) argument may also be relevant in the Russian case. The average (trimmed) payout ratio was 0.40; increasing the payout ratio by 0.10 was associated with an increase in the market-to-book ratio of about 6 percent. Further, we find that widely held firms had higher market-to-book ratios. Though widely held corporations had lower ROA, investors in such corporations were apparently compensated with higher market-to-book ratios. Including share price volatility generates only a weak relationship with corporate market-to-book ratios (perhaps due to small sample size).

VIII

In this article, we document the basic financial structure and dynamics of all industrial corporations in the Russian Empire between 1899 and 1914. We find large differences in capital structures and payout policies across industries, over time, over firms’ life cycles and between ownership structures. These patterns follow predictions of standard corporate finance theory and reflect what we know about the institutions, financial system and process of development in the late Imperial Russian economy. For example, Russian firms’ profits and dividend payments largely followed the business cycle. The relative use of leverage or equity financing was associated with factors like asset tangibility and organizational form in ways consistent with the role of internal agency costs and external information asymmetries. Dividend payout policies appear to have addressed similar issues. While individual corporate founder identities had a measurable impact on funding choices (but not necessarily profitability), the evidence broadly suggests that early industrial corporations could address market imperfections and sustain growth through their interactions with the Imperial Russian financial system. Therefore, constraints on incorporation and on factor and product market development were likely binding considerations for early Russian industrial growth.

Our empirical work relies on a uniquely large and comprehensive panel dataset of corporate financial characteristics in an important historical emerging market. However, there are some important caveats to our results using these data. Although the panel structure, detailed balance sheets, and rich information on corporations’ fixed characteristics substantially improves upon our earlier, cross-sectional work (Gregg and Nafziger Reference GREGG and NAFZIGER2019), we remain hesitant to make fully causal claims given the complicated and simultaneous interconnections between capital structure, governance, payment decisions and profitability. Furthermore, our results describe only industrial corporations, a relatively small subset of all firms in the Russian Empire. However, these were the leading firms of the Empire and were precisely those for which the choices of governance and financing were perhaps most relevant.

Additional research could further illuminate how early Russian industrialization unfolded. For example, available quantitative and qualitative sources may permit a more systematic examination of banking relationships in our context, thereby providing important clues about how Russian corporations interacted with the banking sector. In addition, historical studies of how Russian bankruptcy functioned in practice, or quantitative evaluations of the impact of changes in the tax code, could further reveal how Russian institutions impacted corporate financial outcomes. Each of these topics would benefit greatly from similar studies of early corporate finance in other economies, in order to understand what is specifically Russian and what is more broadly true about capital structure, payout policy and performance in late industrialization.