Causal Effects of the Silver Purchase Program on China

The data and anecdotal accounts of China’s situation in the 1930s reviewed earlier seemingly make it hard to escape the conclusion that US silver pur­chases, and associated pressures for currency appreciation, did significant damage to the Chinese economy.

In this section, the causal relationship between silver prices and Chinese price and exchange rate series is examined through a series of Granger­causality tests. This procedure sheds light on whether past movements in one variable exert significant effects on another. In this way, we can see if silver price movements, for example, played a significant role in accounting for movements in the Chinese variables. We can also see whether the underlying relationships changed, as one would expect, after China made the final break from silver in November 1935. This exercise employs monthly data on the Shanghai wholesale price index, the yuan/$US exchange rate, and the New York silver cash price drawn from the Global Financial Database (http://www.globalfindata.com).

Causality testing reveals that US silver prices Granger-cause Chinese inflation over the January 1928 to October 1935 period but not over the full sample ending in June 1937 (Table 5.6). This is consistent with significant effects of silver price movements on China over the period preceding the November 1935 currency reform. China’s November 1935 break from the silver standard was intended to end its vulnerability to swings in silver prices, and the results are consistent with this goal being achieved. It is, unfortunately, not possible to examine the post-November 1935 period in isolation as there are simply insufficient observations available prior to the July 1937 outbreak of war. Meanwhile, the exchange rate Granger-causes inflation over the whole period, suggesting that exchange rate movements were consistently important over the Nanking decade - and in line with the deflationary effects of currency appreciation discussed in the preceding sections. The conclusions drawn from Table 5.6 are also supported when the equations are reestimated using the real silver price in place of the nominal silver price (Table 5.7).^{[87] [88]}

Silver Shocks and Stock Price Movements

The Granger-causality effects discussed earlier remain limited to the effects of silver prices on nominal variables, owing to the absence of usable series on real variables like output. Although annual estimates by Yeh (1979, p. 97) do suggest that an 8.7% drop in China’s real GDP accompanied the

Table 5.6.

1928:01 -1935:10

1928:01 -1937:06

Notes: ***, **, and * denote significance at the 99%, 95%, and 90% levels, respectively; the figures in brackets are the exact significance levels (P-values); and four lags have been included in each of the test regressions.

onset of US silver purchases in 1934, Brandt and Sargent (1989) and Rawski (1989, 1993) point to Chinese banks’ ability to expand their note issues sufficiently to offset the decline in silver money as world silver prices rose - thereby helping limit (or, much more arguably, eliminate) the negative deflationary effects discerned by Friedman (1992) and many contemporary accounts (see also Burdekin, 2008).

Table 5.7. Granger-Causality with the real silver price in the three-equation system

1928:01 -1935:09

1928:01 -1937:06

Notes: ***, **, and * denote significance at the 99%, 95%, and 90% levels, respectively; the figures in brackets are the exact significance levels (P-values); and four lags have been included in each of the test regressions.

This section focuses on the evolution of the HSBC stock price over the 1927-1937 period, thereby taking advantage of a consistent data series on an institution that stood out as a major player in both Chinese and Hong Kong financial markets. HSBC actually operated as an official arm of Chinese monetary policy prior to the Nanking decade and remained “the most important foreign bank in China [that] had long dominated international financethere... ’’(Schenk,2002). HSBChas,ofcourse,retaineditsposition as a major international bank to this day. HSBC stock traded actively in London and Shanghai as well as in Hong Kong. Although the data employed

Figure 5.5. Hongkong and Shanghai Banking Corporation, January 1927-June 1937 (month-end closing share price in Hong Kong).

in this section are taken from the Hong Kong market, stock prices were generally very similar across markets, reflecting the availability of telegraph communications (Bailey and Bhaopichitr, 2004, p. 146). The HSBC data series spans the period both before and after China’s monetary reform on November 5,1935 that freed the note issue from its old link to the shrinking stock of available silver.^[89]

HSBC’s trading price over the January 1927 to June 1937 period is plotted in Figure 5.5, revealing sustained gains over the late 1920s followed by sharp declines in 1931 and 1934-1935. These declines themselves appear to coincide, first, with the British and Japanese break from gold - coupled with the September 1931 Japanese invasion of Manchuria - and, second, with the sharp rise in silver prices following the implementation of the US silver purchase program. Econometric analysis confirms that each of these declines represented a nonreversed “turning point” in the equity series. The HSBC stock price is initially simply modeled as a function of its own past history, with such series often displaying a certain degree of inertia whereby past movements, on average, help predict future movements.

The results suggest that the effect of the lagged stock price change is positive and significant at the 94% confidence level. Although the overall explanatory power of the basic equation is low, the next step is to allow for additional dummy variables that can capture possible shifts in the equa­tion over the sample period and, hence, better explain the major ups and downs. Following the general procedure suggested by Banerjee, Lumsdaine, and Stock (1992), the data were organized into a series of non-overlapping “windows,” each being a year in length. Repeated estimation, progressively moving through the overall sample period, reveals those windows, or por­tions of the sample, where especially low R² values make the presence of a structural break most likely. Dummy variables are then defined successively for each week in the period surrounding the window in question, with these dummies taking on a value of one for that week and every week thereafter. The week for which this “rolling” estimation procedure yields the highest dummy variable significance level then forms our best estimate of the exact date at which a nonreversed “turning point” occurs.^{[90] [91]}

The most significant breakpoints for the HSBC series are presented in Table 5.8. The results suggest that negative turning points in the series occurred in March and September 1931. The latter downturn occurs in the same month as the United Kingdom’s exit from the gold standard andJapan’s

Table 5.8. Potential TurningPoints in the Hong Kong and Shanghai Bank Corporation Stock Price Series

Notes: The t-statistics reflect the sign and signif­icance of the coefficient attached to a dummy variable defined for the month in question, and **7*, and * denote significance at the 99%, 95%, and 90% confidence levels, respectively.

invasion of Manchuria. The UK policy action gave rise to depreciation of the pound against the Chinese currency whereas the loss of Manchuria deprived China of a key source of industrial production and 15% of its overall customs revenues (Young, 1971, p. 200). The earlier turning point indicated for March 1931 appears to coincide with a correction from the sharp share price advance that began during the summer of 1930, but is not obviously tied to major historical events. Meanwhile, the positive turning point suggested for March 1932 saw HSBC’s stock price begin a partial recovery from the steep downturn initiated in September 1931. The gains continued until 1934 and perhaps reflect the fact that business conditions in Shanghai were not, at first, so adversely affected by the 1931 events, with inflows of silver from China’s interior more than compensating for the outflow abroad (as discussed earlier).

Successive negative turning points are indicated for June 1934 and August 1934. The first negative turning point coincides with the passage of the US Silver Purchase Act on June 19, 1934. Although large-scale purchases did not get underway until the following month, the act clearly presaged new upward pressure on silver prices and hence additional upward pressures on China’s currency. Meanwhile, August 1934 stands out as the month with the most dramatic increase in the silver drain prior to China’s November 1935 break from silver (see Table 5.4). BothJune 1934 and August 1934 were therefore months that contained plenty of bad news for business interests in Shanghai and for the silver-standard economies generally. The specific dates of the negative turning points appear to point, rather unambiguously, to a negative reaction to the initiation, and effects, of the US Silver Purchase Act. HSBC’s stock price certainly dropped precipitously after the act was passed, falling by over 46% between June 1934 and April 1935 - at the very least suggesting that the act was not favorably received by the bank’s investors.

After bottoming out in April 1935, HSBC’s stock price enjoyed a limited “bounce” followed by accelerating gains after the November 1935 currency reform. HSBC’s share price rose by just under 13% between April and October 1935 before embarking on a near 82% advance between October 1935 and June 1937. The significant positive turning point identified for May 1935 coincides with the initial recovery from the low recorded in April 1935 but precedes the larger advance that, based on the raw data displayed in Figure 5.5, appeared to correspond with the November 1935 break from silver. The turning point procedure is, unfortunately, often unable to distinguish breaks in nearby months, and identification is further hampered when proximity to the end of the sample leaves little time for the turning point to be established in the data. The bottom line is that we see evidence that the stock price recovered in 1935, even though the empirical procedure cannot precisely tie the recovery to the actual month of the currency reform.

Overall, the turning point analysis, first, offers empirical confirmation of the negative effects generally attributed to the United Kingdom’s exit from the gold standard and the Japanese invasion of Manchuria. Following par­tial recovery in 1932, renewed successive negative turning points arise just when the US Silver Purchase Act was passed and large-scale silver purchases began. HSBC’s stock plunged after these latter events, and, although there is evidence of a positive turning point as early as May 1935, large gains were postponed until after China’s break from silver later in the year.