RDP 9303: The 1893 Bank Crashes and Monetary Aggregates 2. Existing Estimates of Monetary Aggregates

2.1. Weakness in Data Set

The data provided by Butlin, Hall and White that have been used to generate estimates of the stock of money were estimates of currency in the hands of the public and various categories of deposit liabilities held by trading and savings banks. Problems exist with respect to both the currency and deposit data. Estimates of currency in the hands of the public before 1900 are crude in the extreme and cast doubt on the accuracy of the existing figure of M1, or the stock of money narrowly defined (Butlin, Hall and White, 1971, Table 42, pp. 453–457 and pp. 92–94). Currency was an important component of M1, falling from roughly 40 per cent to less than 20 per cent, over the period from 1890 to 1914. Calculations of the value of notes in the hands of the public, which comprised roughly one third of the total of currency, are reasonably straightforward. A more serious problem arises with estimates of the public's holdings of coin, principally gold. These are not known with any level of precision. The series of currency in the hands of the public generated by Butlin, Hall and White which begins in 1901 has been based on an extrapolation from a series of benchmark estimates of the amounts of gold coin in the hands of the public in 1901, 1906 and 1911 made by the New South Wales statistician in 1921 (Butlin et. al., 1971, p. 93). However, Schedvin (1973, pp. 590–591 and note 6) extended the series back into the nineteenth century by extrapolating on a per capita basis. The manner of estimation of currency prevents any accurate measure of hoarding during the several months of banking panic and bank closures.

More serious errors have emerged from the way in which estimates of broad money have been constructed. The nub of the problem is how to treat the deposit liabilities of the reconstructed banks. Deposit holders, with the exception of governments who enjoyed a privileged position as creditors, were offered three choices. Their deposits at the time of suspension, whether current accounts or interest bearing accounts of three to twelve months term, were converted into some combination of preference shares, debenture stock, some of which was interminable, or deferred deposit receipts whose maturity was spread across anything from three to nine years. Of the more than £60m caught up in the reconstruction schemes, a little over £3m were converted into preference shares and another £3m became interminable stock, leaving some £54m as long dated liabilities. The lengthening of the term from twelve months prior to suspension up to a minimum of four years^[2] before repayment was to begin after reconstruction suggests strongly that these liabilities should be removed from measures of the money stock.

Butlin, Hall and White did make some minor adjustments to several of their deposit liability series as a result of the terms of the bank reconstruction schemes of 1893. They used the quarterly returns to governments by banks of selected liabilities and assets within colonial and later state boundaries as the principal source of their data. Some of these colonies, notably Victoria and Queensland, altered their short form balance sheet categories, namely by introducing ‘other liabilities’ to deal with perpetual stock issued by the English, Scottish and Australian Bank and the Queensland National Bank (Butlin et. al., 1971, Table 12, pp. 140–159). However, these amendments do not inspire confidence as no explanation has been given of how the data for individual colonies, shown in Tables 13–16(i), was calculated (Butlin et. al., 1971, Tables 13–16(i), pp. 160–237). These tables cover all states apart from Tasmania. There is no indication that the English, Scottish and Australian Bank issued interminable deposit stock, redeemable at its discretion, in both its initial and 1895 revised scheme of arrangement with its creditors^[3], or that the Queensland National Bank's interminable stock was issued under its revised scheme of arrangement which took effect in 1897^[4]. Also included in this series of ‘other liabilities’ are the ‘old liabilities of the Commercial Bank of Australia not dealt with under its reconstruction scheme’ for ‘a few years in the 1890s’, those years being 1893 to 1895 (Butlin et. al., 1971, p. 77)^[5]. The remaining extended deposit receipts, which comprise most of the liabilities affected by reconstruction schemes were left in with term deposits of normal maturity (Butlin et. al., 1971, p. 77 and p. 80).

The modest adjustments made by Butlin, Hall and White (shown in Appendix A), are incomplete in several respects. Firstly, the series of ‘other liabilities’ is considerably smaller than the figures for those categories of liabilities supposedly dealt with shown in the balance sheets of the banks in question. Even if the larger balance sheet data are substituted for the Butlin, Hall and White series, they capture no more than a fifth of the deposit liabilities caught up in the reconstruction schemes implemented in 1893, as will be shown below. Secondly, the amounts which have been deducted from the deposits within Australia by Butlin, Hall and White are overstated by roughly a quarter, as they refer to interminable stock which was held in the United Kingdom as well as in Australia. Furthermore, no adjustments to the data of deposit liabilities to allow for the effects of reconstruction appear to have been made to the aggregate balance sheets of trading banks shown in Table 2(i) of Butlin, Hall and White (1971, pp. 83–84 and pp. 118–119).

The data of interminable stock held in the United Kingdom needs to be netted out to give a figure of monetary aggregates within Australia. Schedvin used the data sets in Butlin, Hall and White to pioneer the construction of time series of the money stock, the monetary base and reserve and currency ratios (Schedvin, 1973 and 1972). He made a number of adjustments to the Butlin, Hall and White data in the course of his compilations but none were directly concerned with the problems raised in this paper (Schedvin, 1973, pp. 589–593). Later students of Australian monetary experience, Pope (1986, Table 2, pp. 27–44) and Fane (1988, Statistical Appendix), have accepted the Schedvin and Butlin, Hall and White material relating to deposit liabilities held within Australia through the period under review. However, it should be noted that the series of M2 generated by Pope is below that of Schedvin by a margin of 3 per cent rising to around 7 per cent over the period under consideration. This divergence results from Pope's netting out of deposits held in trading banks by savings banks in his calculation of M2 whereas Schedvin makes this adjustment for M3 (Pope, 1986, pp. 27–28 and Schedvin, 1973, pp. 592–593). This elimination of the double counting of savings bank deposits does not affect the figure of M3 which is virtually identical in both series.

2.2. Measurements in Monetary Aggregates 1893–1914

The Schedvin series, presented in Table 1, suggests that a severe contraction in the stock of money occurred in the 1890s. In nominal terms M1 and M2 both contracted around 22 per cent from peak to trough. However, the various aggregates did not follow a uniform path with respect to the timing or the extent of their movements. As Table 1 shows, M1 peaked in 1888 and reached its nadir in 1893, while M2 peaked later in 1890 and reached its minimum value in 1898 after a marginal upward movement in 1895 and 1896. On the other hand, M3 peaked in 1892 before shrinking by 11 per cent to a trough in 1894, a figure matched again in 1898. Recovery to the previous peak was slow for all monetary aggregates. In nominal terms M1 surpassed its earlier peak figure in 1896, M2 in 1906 and M3 in 1900. The more rapid recovery of the latter resulted from the flight of depositors towards state-owned savings banks whose liabilities were guaranteed by colonial governments.

These data suggest that the contraction in two of the series of monetary aggregates, M1 and M2, predate the bank crashes of 1893. Only M3 peaked immediately before the banking crisis. Why then did monetary aggregates decline? Schedvin offers an explanation of the movements in monetary aggregates with reference to a Cagan-style analysis of changes in high powered money, or the monetary base, and ratios of currency and cash reserves to deposits (Schedvin, 1973, pp. 593–598 and 1972 and Cagan, 1965). The monetary base shrank quickly from £15.930m in 1887 to £3.648m in 1892, while its effect on the money stock was ameliorated to a degree by sharp falls in the reserve ratio from 7.38 per cent in 1887 to −5.22 per cent in 1893. This shift in the reserve ratio was reinforced by a marginal reduction in the currency ratio (Schedvin, 1973, pp. 601–605 and 1972).

Such an explanation clearly suggests that the bank suspensions in 1893 were not the primary cause of the contraction in monetary aggregates. The underlying cause, the shrinkage of the monetary base, predates the bank crashes by a number of years. However, there was a larger decline in all categories of the money stock, M1, M2 and M3, in 1893 than in any other single year. M1 fell nearly 11 per cent, M2 nearly 14 per cent and M3 marginally more than 10 per cent. The thrust of this paper is to argue that the bank crashes had an even more significant impact on broad money aggregates than these annual movements suggest.

Footnotes

Under the original schemes of arrangement no repayments were due before 1896 and the Bank of North Queensland was to be the first institution to complete repayment by discharging its obligations in 1899. Most of the repayments were to fall due between 1898 and 1901 (Banker's Magazine (1894) p. 890). [2]

See Australasian Insurance and Banking Record (hereafter AIBR) (1893), pp. 606–607 and (1896), pp. 417–418. [3]

See AIBR (1897), pp. 544–545. [4]

The correct amounts are shown in Appendix A. [5]