RDP 1999-06: Two Depressions, One Banking Collapse 3. A Comparison of Indicators of Financial System Stability

The central thesis of this paper is that the variation in the performance of the financial system across the 1890s and 1930s stems mainly from differences in the condition of the financial systems that were evident well before the economic downturn in each episode. This is demonstrated by comparing six broad indicators of financial system stability across the decade or so prior to each depression.

We refer to financial instability in terms of the ex ante probability of a financial system disturbance of sufficient size that it implies noticeable macroeconomic effects (Kent and Debelle 1999).^[24] In this paper we focus on indicators that relate mostly to the degree of credit risk in the financial system, although we also discuss the related issue of liquidity risk. The indicators we focus on include:

1. the level and nature of investment;
2. property market speculation;
3. credit growth;
4. capital inflows;
5. degree of risk management within the financial system; and
6. competitive pressures in the financial sector.

These indicators are of course closely related, and it is not clear that there exists an obvious ranking of their importance for system stability.^[25] Arguably the level of credit and the speed at which credit is expanding are the key factors behind many episodes of financial instability. Even so, we choose investment as a starting point for the discussion since it relates to many other indicators and points to the source of the initial positive shock that triggered the expansion and subsequent instability.

3.1 Investment – Public and Private

Total national investment grew strongly leading up to both depressions, and then fell dramatically at the outset of each depression. As a share of GDP, investment was not that much higher through the 1880s compared with the 1920s – averaging around 18 per cent in both decades. However, investment had remained relatively high over a longer period leading up to the 1890s.^[26] It had been above 15 per cent of GDP for 17 consecutive years to 1891, whereas it had been above this level over only 10 consecutive years to 1929. Moreover, what distinguished the 1880s from the 1920s was the composition of investment, in terms of both the split between public and private investment, and the nature of this investment (Figure 8). These compositional differences had implications for the relative stability of the financial system over these episodes.

Private investment as a share of GDP was sustained at a higher level, and over a longer period, prior to the 1890s depression than it was prior to the 1930s depression. Although highly volatile, private investment averaged about 11 per cent of GDP from 1875 to 1891, compared with an average of just under 9 per cent from 1920 to 1930. Investment reached a seven-year high of about 13 per cent of GDP in 1888, coinciding with the peak of the property price boom (see below). Through both depressions investment fell rapidly, though the fall in the 1890s was larger and lasted longer than was the case during the 1930s.

Public investment spending also grew strongly leading up to both depressions, but was lower on average during the 1880s than during the 1920s – 7 per cent compared with about 9 per cent of GDP.

These differences between the two episodes affected financial stability in a number of ways. The financial system will become less stable if a substantial proportion of lending by financial institutions is used to fund increasingly riskier investment projects. This was more likely during the 1880s than the 1920s because aggregate investment had been sustained at a high level for longer, and private investment was a much larger share of aggregate investment. It is plausible that relatively high levels of investment, if sustained over a very long period of time, will lead to increasingly riskier projects being undertaken, especially when investment is concentrated in only a few sectors of the economy (as it was during the 1880s). Also, a higher share of private investment implies more exposure for financial institutions to the risks inherent in these projects. This follows from the fact that private investment was (and still is) more reliant on financial intermediation than government investment. Furthermore, a greater share of public investment should directly reduce the volatility of aggregate demand since governments are generally better placed to fund investment projects even during downturns.^[27]

Figure 8 clearly shows that compared with public investment, private investment fell earlier and further, and stayed lower for longer during both depressions. Also, the cycle in private investment was strongly correlated with the cycle in bank credit during the 1880s and 1890s. However, this relationship was not nearly as strong over the 1920s and 1930s.

The investment boom of the 1870s and 1880s was driven by a number of factors. Extremely high rates of population growth during the 1870s and 1880s helped to drive rapid real GDP growth – output was expanding rapidly through the application of imported capital and labour to the development of many previously unexploited resources and investment opportunities. However, productivity growth was not especially strong. That is, output per capita grew consistently, though not that rapidly – from 1861 to 1891, GDP per capita grew on average about 1 per cent per annum, compared with an average of almost 3 per cent per annum in the 30 years following the Second World War. A very high demand for housing was driven by population growth and bolstered by wealth accumulated in the agricultural sector and in the gold fields flowing back into the major cities. Underlying these developments in the private sector, governments were increasing spending on infrastructure such as railways and communications.

Therefore, it is not surprising that investment in the 1870s and 1880s was dominated by construction activity. The strength of construction over this period cannot be overemphasised since it represented the biggest building boom in Australia's history (Figure 8). Much of it was concentrated in urban centres, especially Melbourne which was undergoing rapid expansion and was the focal point for speculation in the property market which eventually spread to other colonies (Boehm 1971). From 1875 to 1891, building activity as a share of GDP averaged around 14 per cent, compared with an average of only 9 per cent from 1920 to 1930.^[28] It would not be an overstatement to claim that this level of activity over the 17 years to 1891 represented the most extravagant of building booms.

Even though population growth provided a fundamental reason for the construction boom of the 1870s and 1880s, building activity remained high even after the rate of population growth slowed markedly towards the end of the 1880s.^[29] This by itself was a source of instability.

Population growth in the 1920s was considerably slower than during the 1870s and 1880s. Hence, the pressure on the property market was nowhere near as great – construction activity did increase, though a greater proportion of this was public. During the 1920s, public construction accounted for slightly more than half of building activity (Butlin 1962). In contrast, private activity accounted for on average 60 per cent of the construction during the boom years of the 1870s and 1880s.

In summary, one of the major differences between the two depression episodes was the unprecedented building boom prior to the 1890s depression. A large proportion of this activity was undertaken by the private sector. The dramatic boom and bust in private investment, building activity and bank credit were all closely related over the 1880s and 1890s. Cycles in these indicators were comparatively muted through the 1920s and 1930s.

3.2 Speculation in the Property Market

During both episodes, rising asset prices, especially for property, went hand in hand with increasing investment and building activity. Downturns in asset prices, investment and building activity were just as closely related.^[30] Melbourne was the centre of a boom in property prices that reached a peak in 1888 (Boehm 1971). Boehm highlights a number of factors driving this boom, including strong population growth, particularly among the working age population and urbanisation.^[31] The land boom was supported by the large number of building societies that opened (Figure 5) and the view that one couldn't lose money by investing in land (Cannon 1966). Legislation covering building societies was changed in 1876 to allow them to buy and sell land themselves. This resulted in building societies becoming little more than ‘speculative operations’ which added to the inflationary pressure on land and property values. Although an accurate time series of property price data is unavailable, Silberberg (1975) presents data suggesting that the average net nominal annual rate of return on land in Melbourne was about 35 per cent from 1880 to 1892. No comparable studies are available for the 1920s. Cannon (1966) cites anecdotal evidence such as a city block in Melbourne almost doubling in value in a couple of months in late 1887, while Daly (1982) estimates that the average price of land in Sydney increased by over 80 per cent from 1880 to 1884. Land prices also increased rapidly in the first half of the 1920s, but Daly suggests prices remained relatively stable leading into the depression.

In order to provide a rough approximation to movements in property prices we attempted to exploit data on the aggregate capital value of ratable properties in Melbourne and Sydney. Movements in this measure reflect changes in prices, as well as changes in the volume and quality of properties – both of which we would expect to be on an upward trend or at least not likely to decline rapidly. Figure 9 compares aggregate capital values over the two depression episodes for the cities of Melbourne and Sydney.^[32] Figure 9 suggests that price rises over both boom periods and in both cities were of similar orders of magnitude.^[33] However, this finding is at odds with contemporary and historical accounts of the two cycles.

One reason why these data might not fully reflect price rises is that capital values were determined for the purposes of tax assessments by the valuation departments of city councils. Councils generally adopted a conservative approach by attempting to value property according to the underlying or ‘real’ value of property as opposed to the current market value of property. If this process was done methodically and consistently for both episodes then the capital values should provide a reasonable indication of the relative size of the cycles in property prices over the two episodes. However, the Sydney Morning Herald (1930) suggests that, at least for Sydney, the practice of more conservative council evaluations had been ignored during the boom leading up to the 1930s and that valuations were influenced by a ‘…number of spectacular sales of city properties’. Thus, while capital values of the 1880s may have underestimated actual price rises, this bias seems to have been less prevalent in the 1920s.

Despite the paucity of data, anecdotal evidence and contemporary accounts suggest that speculation in property markets was much more acute during the 1880s than the 1920s. This conclusion is supported by a comparison of building activity across these two decades. The combination of extraordinary high levels of construction and substantial speculation in property markets were important factors behind increasing vulnerability of the financial system through the 1880s. This effect was reinforced by the cycle in credit.

3.3 Credit

Historically one of the major causes of financial instability has been rapid increases in bank lending in conjunction with unsustainable rises in asset prices. In recent times the catalyst for this process has been financial deregulation. In a way, this force was also present in the 1880s. The Australian banking system in the 1800s had operated under direction of the British Treasury in accordance to the ‘real bills doctrine’, which among other things prohibited lending backed by land.^[34] However, Pope (1991) suggests that after the arrival of ‘responsible government’ in the 1850s, these regulations were largely ignored and banks increasingly engaged in lending for speculative purposes. Increasing willingness to ignore the real bills doctrine was at least in part a response to increasing competition from non-bank financial institutions (Section 3.6).

The result of these changes was a dramatic increase in the ratio of bank credit to GDP (Figure 10). This rise was sizeable compared with the experience of the 1920s. Although data on the direction of lending is not readily available over the earlier episode,^[35] contemporary accounts suggest that a large proportion of lending in the 1880s was directed to property speculation and development. This is consistent with the very high level of building activity (Figure 8), and the fact that the cycle in building activity coincided with cycles in credit and property prices.

3.4 Banks' Balance Sheets and Foreign Borrowing

The rapid expansion of banks' balance sheets contributed to the instability of the banking system leading up to the crash in 1893. The rise in credit over the 1880s was so dramatic that banks were unable to fund their lending through increases in domestic deposits. Bank advances as a share of deposits rose over the 1880s, reversing an earlier downward trend (Figure 11). Banks were able to make advances in excess of their Australian deposits by seeking deposits in Britain, leaving the banking system exposed to external developments. This is consistent with the substantial increase in capital inflows over this period. Capital inflow was above 6 per cent of GDP for eight consecutive years during the 1880s; it breached this level in only two years during the 1920s.^[36]

Through the 1920s, trading banks expanded their lending, but at a rate more in line with increases in Australian deposits. The greater conservatism of the banking sector, particularly with regard to credit policy, was an important factor in providing a more stable financial system leading into the 1930s depression.

The proportion of private sector borrowing to total borrowing was higher in the 1880s than the 1920s (Schedvin 1970).^[37] This may have insulated the financial system from overseas developments in the 1930s relative to the 1890s for two reasons. First, the public sector is generally perceived to represent a lower credit risk than the private sector.^[38] Second, when funds are channelled through the banking system, in addition to the risk of default by the final user of the credit, there is also the risk of default by the intermediary. From 1886 to 1891, Australian bank liabilities to British residents averaged 45 per cent of liabilities within Australia. This ratio averaged only 15 per cent from 1920 to 1927 (Butlin, Hall and White 1971).

The cessation of capital flows in both episodes was due in part to financial turmoil in the US and UK. The majority of overseas borrowing during the 1880s was done in London. The Barings crisis in 1890 led to a reduction in the availability of British credit. Barings was a London discount house which faced liquidity difficulties in mid 1890 due to its exposures in South America. The default on these loans brought attention to all overseas securities and Australian governments found it difficult to raise new loans in Britain (Boehm 1971). Because governments encountered difficulties raising new loans, the Australian private sector also found it more difficult to borrow money in Britain.

Schedvin (1970) suggests that some loans were obtained from New York during the 1920s, but the majority of borrowing was still done in London. The London market was disrupted in the late 1920s by speculative activity in Europe and large outflows associated with the stock market boom in New York. London was virtually cut off as a source of funds for Australian long-term borrowing in early 1929 (Royal Commission 1937, paragraph 114).

In summary, the build up in credit during the 1880s was more substantial than in the 1920s and it was also more reliant on large capital inflows, much of this in the form of private borrowing. This increased the instability of the financial system by increasing its vulnerability to foreign financial shocks.

3.5 Risk Management – Prudence and Diversification

The degree of prudence of the banking system declined over the 1880s. Trading banks were increasing their risk – as measured, for example, by the ratio of lending to deposits – at the same time as reducing liquidity. One measure of liquidity is the ratio of trading bank cash balances to deposits within Australia (Figure 12). Cash balances consisted of coin, bullion and Australian notes.

This measure of liquidity actually behaved similarly in the periods prior to each depression. The trading banks' cash-to-deposits ratio fell from the relatively high levels of the 1870s to a trough of 13.8 per cent in 1883, before picking up and stabilising until the onset of the depression. Despite a sharp fall in deposits in 1893, trading banks were able to increase their cash balances and hence, liquidity rose sharply.^[39] They were able to do this by reducing advances to such an extent as to more than offset the fall in deposits and by raising new capital. The cash-to-deposits ratio fell over the 1920s. In 1929 the ratio was around 17 per cent – a similar level to the years immediately prior to the 1890s episode. Indeed, Schedvin (1970) suggests that the banking system experienced a severe liquidity crisis towards the end of 1929.

When comparing the ability of the banking system to withstand large shocks, it is also useful to examine a broader measure of the soundness of the system. When trading banks' holdings of government and municipal securities are added to cash balances, the ratio to deposits averaged around 40 per cent over the second half of the 1920s – double the average ratio over the second half of the 1880s. This enabled the banking system to better withstand the macroeconomic downturn and subsequent increases in loan defaults.

Declining levels of prudence over the 1880s are also reflected in a decline in the ratio of paid up capital to assets, while retained earnings as a proportion of assets remained flat (Figure 13). This was due to a rapid increase in assets as well as relatively high dividend rates, averaging 8.3 per cent of shareholders' funds over the 1880s.

In contrast, through the 1920s trading banks increased paid up capital and retained earnings at a faster rate than total assets, reflecting the greater conservatism in the banking system. The ratio of shareholders' funds to assets was increased to nearly 20 per cent in 1929, from 13 per cent ten years earlier. During the 1930s, the trading banks were able to draw on these funds.

With lower capitalisation of the banks during the early 1890s relative to the late 1920s and early 1930s there was a greater probability of a given internal or external shock leading to bank insolvency. Hence, variation in bank capitalisation was an important explanator of variation in financial system stability.

Increased geographical diversification of lending within a given bank tends to increase the stability of the financial system by reducing banks' exposures to specific regions or industries. Of the 23 trading banks operating at the start of the 1890s depression, only ten were in existence at the start of the 1930s depression. Each of these ten banks increased the share of assets held outside their home state between the two depressions (see Appendix B for details). In fact, banks were tending to spread their lending more evenly across all of the six Australian States and New Zealand. This is demonstrated by examining each bank's distribution of the share of their advances across these seven regions. Figure 14 shows that from 1891 to 1929 there was an almost universal reduction in the variance of these distributions.^[40]

Another important factor affecting risk is the extent of management control over lending practices. It would appear that this control was increasingly lax over the 1880s. In part, this was due to the rapid expansion of branch networks. Butlin, S.J. (1977; 1986) shows that the number of trading bank branches rose from about 800 in the late 1870s to a peak of 1,534 in 1892.^[41] Through the depression, trading bank branch numbers fell to 1,235 by 1896, and grew thereafter, although at a more moderate pace than before.

Rapid expansion of branch numbers can become a problem for an individual bank if there is a loss of internal control. This view is emphasised by Merrett (1989) who further suggests that the demand for competent bankers exceeded supply as branch networks expanded. Some evidence of over expansion of branches is provided by comparing the behaviour of those banks that suspended payment in 1893 with those that did not, in terms of the rate of branch expansion over the 1880s. The average bank that suspended payment in 1893 had added 53 branches to its network from 1880 to 1891 compared with only 21 branches for the average bank that did not suspend payment in 1893.^[42] These figures represent average growth rates of about 150 per cent and 90 per cent respectively for the suspending and non-suspending banks. Interestingly, the banks that suspended payment in 1893 were also larger in 1880 than the average non-failing bank – 52 branches compared with 32 branches respectively.

In contrast, it seems that the pressure on banks to expand their branch networks had eased following consolidation of the trading banks after the First World War, and the lack of competition from non-bank financial institutions. Trading bank branches grew on average by about 43 per cent between 1921 and 1930.

An additional factor in increasing the instability of the banking sector in the 1880s was the widening mismatch of the maturity profiles of assets and liabilities (Merrett 1989). Banks began to make a greater proportion of loans on a longer-term basis, while the majority of liabilities remained as shorter-term deposits. The result was a decline in the liquidity of banks' balance sheets, at a time when loan repayments began to decline.

3.6 Competitive Pressures

During the 1880s, banks faced increasing competitive pressure from the emergence of a large number of non-bank financial institutions. This led to banks increasing their level of risk in order to maintain returns, and was one of the important catalysts for the excessive expansion of their balance sheets.

One of the primary aims of these newer institutions was the provision of finance for property speculation. Banks which had been competing on a non-price basis through avenues such as increasing branch numbers were now forced to compete for market share with the new institutions. To achieve this, banks lowered their credit standards and became involved in speculative activities, including loans to the new land finance companies (Boehm 1971).

By contrast, leading up to the 1930s depression, although the trading banks had lost market share, it was mainly to the more conservative savings banks. Building societies and other non-bank financial institutions failed to regain the market share they had lost in the late 1880s and early 1890s (Figure 5).

Also, within the set of trading banks there had been considerable consolidation between the two depressions. Between 1917 and 1927 there were 11 amalgamations between the trading banks, leaving 10 at the onset of the 1930s depression. There was one further amalgamation in 1931. This move towards consolidation may have lessened competition in the banking sector at the same time as providing a vehicle for banks to diversify geographically.

Almost all of the indicators we have examined imply a dramatic increase in the vulnerability of the financial system through the 1880s and early 1890s to adverse macroeconomic shocks. Private investment was relatively high, reflecting an extraordinary building boom which helped to ignite, and was then extended by, speculation in the property market. Simultaneously, there was a massive build up in the level of intermediated finance, supported by large private capital inflows. Increasing risk in the financial system came hand-in-hand with reductions in the capital ratios of banks. Banks' greater appetite for risk was driven by competitive pressures arising from a rash of new non-bank financial institutions. Non-banks and (many, though not all) trading banks became all too willing to lend for speculative purposes using a seemingly inexhaustible supply of funds from overseas.

While many of these pressures were present during the 1920s, they were smaller in magnitude. Private investment was lower, as was building activity. It appears that speculation in property markets was less vigorous. A rise in credit came late in the decade and was relatively minor, as were capital inflows. At the same time there were a number of positive developments for financial stability. Capital ratios of banks increased strongly over the 1920s and banks were more diversified geographically than they had been during the 1880s. Finally, a notable absence of competition from non-bank financial institutions and earlier consolidation among the trading banks helped to foster more prudent lending by banks.

Determining the cause of the greater conservatism in the banking industry during the 1920s compared with the 1880s is beyond the scope of this paper.^[43] Instead, we make three brief observations. First, amalgamation within the set of trading banks and the earlier failure of ‘fringe’ bank and non-bank financial institutions helped to produce a less competitive environment. Second, the greater conservatism in the 1920s may have been due in part to the memory of the disaster of the 1890s.^[44] Third, the conservatism of the banking system through the 1920s appears to owe nothing to the role of the Commonwealth Bank. The Commonwealth Bank played no active part in setting prudential standards (Schedvin 1992). Indeed, the Commonwealth Bank did not regard itself to be a central bank in any effective sense until as late as 1929 (Commonwealth Bank of Australia 1936).

Footnotes

For a recent discussion of the issues related to defining system stability see Crockett (1997). Bernanke and Gertler (1990) present a model of the relationship between financial fragility and performance in the investment sector and the economy overall. They define financial stability as depending on the net worth of potential borrowers. Mishkin (1997) describes financial instability as occurring when information flows are disrupted to such an extent that the financial system cannot efficiently channel funds to productive investment projects. [24]

There are potentially many other indicators of financial instability. For example, a credit financed consumption boom, an overvalued real exchange rate, a rapid increase in interest rates (particularly in foreign financial markets which are the source of capital flows) or inappropriate fiscal and monetary policies, to name a few. We discuss some of these later in the paper. [25]

The level of investment in the late nineteenth and early twentieth centuries was quite low as a share of GDP compared with the post-Second World War average of around 25 per cent (Edey and Britten-Jones 1990). However, it is hard to make comparisons across these eras due to disparities in the level of development and the comparability of data. [26]

However, governments were less able/willing to raise taxes during the 1890s depression and had relied on funds from London to fund investment over the 1880s (see below). [27]

During the building booms of the early 1970s and late 1980s the share of construction in GDP was less than 9 per cent and 8 per cent respectively. [28]

However, there was considerable variation across the colonies in terms of the timing of the building cycles and changes in rates of population growth (Boehm 1971). [29]

Kiyotaki and Moore (1997) present a theoretical model based on such a relationship. For a description of the relationship between property prices, credit and output in Australia during the property booms of the 1970s and 1980s see Kent and Lowe (1997). [30]

Victoria's population increased by 278,000 (or 32 per cent) from 1881 to 1891. By absorbing about 75 per cent of this increase, Melbourne's population grew by over 70 per cent. [31]

Given that the regions defined by the cities of Melbourne and Sydney were relatively small and already well developed (even by the early 1880s), it is reasonable to assume that increases in volumes would have been limited. Other things being equal, ongoing quality improvements imply that rising capital values of property will overstate true price rises. [32]

However, the fall in capital values in Melbourne during the 1890s downturn was particularly large, suggesting ex post that the initial rise was especially unsustainable. [33]

Prior to the 1850s, the British Treasury had laid down regulations for colonial banks governing aspects of banking such as debt-to-capital ratios, loans to directors and frequency of statistical returns. For Australia, perhaps the most significant regulation was that which prevented property being taken as security against advances (Butlin 1986; Pope 1991). [34]

Data for selected years and (five) selected trading banks are available around the 1930s depression (Butlin, Hall and White 1971). Advances allocated specifically for building purposes among these banks represented around 4 per cent of their total advances in 1927 and 1935 (although, this fell to 3.6 per cent in 1930). [35]

Capital flows in Figure 11 prior to 1900 are based on indirect estimates from Butlin (1962). While a range of other estimates of capital flows are available (for example, direct estimates from Butlin 1962 and indirect estimates from Boehm 1971) they display broadly similar movements, especially over the 1880s and early 1890s. [36]

This is also consistent with the pattern of investment and building activity. [37]

However, the default on debt by the NSW State Government during the 1930s depression may have shaken this belief. [38]

While cash balances in the trading banks increased from 1892 to 1893, they fell by about 6 per cent from March to June 1893 (based on seasonally adjusted quarterly averages). From March to June 1893 the ratio of cash balances to deposits remained at about 19 per cent. [39]

For simplicity we have chosen to focus only on the geographical dispersion of lending. We did not attempt to systematically account for differences in the size of the economies of the seven regions. However, a casual examination of Table B1 (Appendix B) suggests that for most banks in the sample, the shares of lending across regions was more closely correlated with the economic size of regions in 1929 than it was in 1891. [40]

Certainly, compared with Britain, Australians were grossly oversupplied with bank branches even during the early 1880s (Pope 1987). In England and Wales in 1880 for every bank branch there were about 12,000 persons. The comparable number for Scotland was 4,000 persons, and for Australia, only 2,500 persons. By 1891 this had fallen to around 2,000 persons. The geographical dispersion of economic activity may have justified branch numbers in Australia being relatively higher than in the United Kingdom. However, much of the Australian population was contained in urban centres, even during the mid to late nineteenth century (Boehm 1971). [41]

There is a fair degree of variation around these averages within the two banking groups, but the conclusions are robust to the exclusion of the outlying observations. [42]

See discussion in Schedvin (1970). [43]

Butlin and Boyce (1985) suggest that during the early 1930s, first-hand experience of the 1890s depression focused bankers' attention on the need for sound financial management. There is some evidence in the 1920s and early 1930s of references regarding the experience of the 1890s – for example, see The Australasian Insurance and Banking Record (AD3R) (1930), Vol. LIV, No. 9, p. 777. Also, a history of the Commercial Banking Company of Sydney shows that some very senior staff of the 1880s and 1890s still held senior positions through the 1920s and early 1930s (Commercial Banking Company of Sydney 1934). In fact, George Judah Cohen was a director of the bank from 1885 to 1933, and held the position of chairman in 1892 and then again from 1900 to 1933. [44]