RDP 9513: Asset-price Bubbles and Monetary Policy 3. What Can Economists Say About Financial Market Volatility?

In seeking to understand volatility in bond and foreign exchange markets, it is of interest to touch on the wider debate about financial market volatility. There has been a lively academic debate, given initial impetus by Shiller (1981) and LeRoy and Porter (1981) about whether financial market volatility is ‘excessive’ or not. The debate focuses primarily on the stockmarket and on the issue of whether the volatility of stock prices can be justified by the volatility of the discounted stream of future dividends. Ultimately, the relevant statistical tests have a joint null hypothesis of market efficiency and a specific model of the discount rate used to discount future dividends.^[11] As a consequence, when the data imply rejection of this joint hypothesis (as they invariably do) it is not clear whether this is a demonstration that financial market volatility is indeed excessive, compared to the volatility to be expected of an efficient market, or instead, simply a rejection of the specific model of the discount rate (see Shiller (1989), and comments on Shiller by Cochrane (1991)).

There is, however, other evidence about the nature of asset market volatility provided by two ‘events’ in the stockmarket. Although not new, this evidence is compelling and hence worth examining. The first event is a paperwork backlog (!) at the New York and American Stock Exchanges, which led these exchanges to be closed on Wednesdays during the second half of 1968.

French and Roll (1986) use this event to compare the movement of stock prices from the Tuesday close of the exchange to the Thursday close in weeks when the exchange was closed on Wednesday because of the paperwork backlog, with the movement in weeks when it was open. Paperwork backlogs at the stock exchange should be irrelevant to the Tuesday-close-to-Thursday-close performance of companies listed on the exchange. Hence, if stock prices move solely because of the arrival of new relevant information about the companies listed, then the average variance of stock returns in a two-day period including a Wednesday exchange holiday should be the same as an average two-day period with the exchange open on both days, or equivalently, twice the variance of an average single day on which the exchange is open.^[12] In fact, French and Roll find that the average variance of stock prices over two days including an exchange holiday is much closer to the variance over an average single day than an average two-day period with the exchange open on both days.^[13]

The second event that casts light on asset market volatility is the 1987 stockmarket crash. Based on questionnaires completed in its aftermath by both institutional and individual investors, Shiller (1988) concludes that:

no news event, other than news of the crash itself, precipitated the crash. Rather, the dynamics of stock market prices seem to have more to do with the internal dynamics of investor thinking, and the medium of communications among large groups of investors is price. In a period when there is a widespread opinion that the market is under or overpriced, investors are standing ready to sell. It takes only a nudge in prices, something to get them reacting, to set off a major market move.

Clearly, neither of these examples imply that asset prices do not respond to changes in economic fundamentals. They do, however, provide compelling evidence that some of the short-term movement in asset markets cannot be explained in terms of the efficient incorporation of public information about fundamentals. Instead, at least some asset price volatility appears to arise from the process of trading introducing noise into asset prices.^[14]

Returning to volatility in the bond and foreign exchange markets, it is worth commenting on economists' different level of understanding of these two markets. In the bond market, there is little controversy about the determinants of bond yields. There is a simple underlying model of nominal bond yields and agreement among economists about the explanatory power of this model. As we have discussed, the nominal bond yield can usefully be decomposed into the expected real yield and expected inflation over the life of the bond. Although risk premia differ between countries, expected real yields on government long bonds are similar in OECD countries with open capital markets and infinitesimal risks of default. Furthermore, expected future inflation responds, probably with a lag, to actual inflation, so that differences in actual inflation explain a substantial part of differences in nominal bond yields between countries (Tables 1 and 2). Although the bond market moves in puzzling ways at times, with 1994–95 being a prime example, economists are rightly confident that they have a good understanding of the economic forces that determine bond yields.

Unfortunately, the same cannot be said of the foreign exchange market. For OECD countries with moderate inflation rates, it is true that PPP provides some guide for movements in floating exchange rates over many years (Table 6). Over shorter periods of time, however, there is simply no underlying model, agreed upon by economists, that explains the movement of exchange rates. Instead, exchange rates are apparently subject to a myriad of influences, and there has been little success uncovering the economic fundamentals – or, for that matter, other forces – that determine their shorter-term movements. As Richard Meese (1990) puts it:

The proportion of (monthly or quarterly) exchange rate changes that current models can explain is essentially zero. Even after-the-fact forecasts that use actual values (instead of forecasted values) of the explanatory variables [italics added] cannot explain major currency movements over the post-Bretton Woods era. This result is quite surprising.

The extent to which fundamentals explain the shorter-term movements of bond yields and exchange rates is relevant to understanding volatility in these two markets. In the bond market, where economic fundamentals provide a convincing explanation for much of the movement of bond yields, one might reasonably expect a change in economic fundamentals – like a fall in the volatility of inflation – to have a significant and predictable influence on bond yield volatility. By contrast, in the foreign exchange market, where for reasons that are not fully understood, economic fundamentals apparently explain very little of the movement of exchange rates over times of relevance to volatility, one should be less confident that changes in economic fundamentals will have a measurable influence on market volatility.

These observations accord quite well with our empirical results. The world-wide fall in the 1990s in the volatility of inflation seems to have been responsible for at least some fall in the volatility of bond yields. By contrast, and notwithstanding the predictions of economic theory, there has been little, if any, fall in the volatility of exchange rates despite a substantial fall in the volatility of inflation differentials between countries.

Footnotes

For example, two common specific models are that the discount rate is constant through time, or that it is equal to the real interest rate plus a constant risk premium. [11]

With stock price movements closely approximating a random walk, the average variance over two days is twice the average variance over a single day. [12]

The average two-day variance spanning an exchange holiday is 14.5 per cent higher than an average single ‘open’ day, whereas an average two-day period with the exchange open on both days has a variance of stock price movements 75 per cent higher than a two-day period spanning an exchange holiday. [13]

It is beyond our scope to discuss the social costs of excessive financial market volatility. Even if there is substantial volatility introduced by the process of trading, however, the associated social costs may be small (Cochrane 1991, pp. 20–23). [14]