FRBSF Economic Letter
1996-36 | December 13, 1996
On a number of occasions this year, the Labor Department has released reports showing that employment was growing more rapidly than analysts had anticipated. For example, the economy added roughly 800,000 jobs in February and another 239,000 jobs in June. Many people regarded this as good news, as it was a sign of better prospects for job seekers and of greater security for those already at work. Yet the Dow Jones Industrial Average fell by 3 percent in response to the February employment report and by 2 percent in response to the June report. Some observers were puzzled by this reaction and wondered why stock prices would fall in response to news about strong economic growth. This Economic Letter offers one explanation.
Stock prices are influenced by two countervailing forces. On the one hand, firms pay dividends to their shareholders, and when there is good economic news investors may expect these payments to increase. Other things equal, this would tend to raise stock prices. On the other hand, a promise to pay $1 sometime in the future isn’t worth $1 today. Borrowers generally have to pay something extra to lenders in order to induce them to part with their money, and this means that the current price of a promise to pay $1 sometime in the future is less than $1. In other words, promises of future payments sell at a discount. Good economic news tends to increase the rate at which investors discount future payments, so holding expected dividends constant, it would tend to depress stock prices. The net effect on stock prices depends on the relative strength of these two forces. Stock prices rise if the dividend effect is greater, and they fall if the discount effect dominates.
Why does good economic news tend to increase the rate at which investors discount future payments? In an efficient capital market, the expected nominal discount rate is (approximately) equal to the interest rate on a nominally default-free riskless bond. To see why, suppose that borrowers were selling claims to a sure $1 payment one period hence. If the yield on this claim were higher than the rate at which savers discounted the future, there would be an excess demand for this security, and this would bid up its price and reduce its yield. Hence market forces tend to equalize nominal discount rates and risk-free interest rates. Treasury bills are nominally riskless, and their yields tend to rise in response to strong economic news. This means that expected nominal discount rates also tend to rise in response to this news.
The discount effect is illustrated in its purest form by the reaction of prices on long-term Treasury bonds to news about strong economic growth. These securities have fixed coupon streams and fixed payments at maturity, and their prices are determined by discounting these payments back to the present. Positive news about the economy has no effect on the stream of payments, because they are predetermined, but it does affect the rate at which those payments are discounted. In particular, since the Federal Reserve tries to lean against the wind, it is more likely to raise the short-term interest rate and less likely to lower it when the economy is strong. Thus, when investors learn that the economy is stronger than anticipated, they revise upward their short-term interest rate forecasts, and long-term bond prices fall because the predetermined flow of payments is discounted at higher rates.
How are equities different? Among other things, one difference between equities and bonds is that the flow of dividends is not predetermined. Firms can increase or decrease dividends in response to changes in earnings, and that is why some people expect stock prices to respond differently to strong economic news. For example, the conventional wisdom goes as follows. Presumably, firms are hiring more workers because they are forecasting higher demand for their goods and services and want to increase production. But if that is the case, an increase in employment should be a sign that earnings are likely to rise. And since stock prices reflect the present discounted value of future dividend payments, stock prices also should increase.
This argument implicitly rests on two assumptions: first, that dividends rise and fall with earnings over the business cycle and, second, that the market discount factor is either constant or does not vary much over the business cycle. Neither of these premises is valid. While firms do tend to increase dividends in response to a permanent increase in earnings, they usually try to smooth over transitory changes. There is also much evidence that the market discount factor is highly variable. Taken together, these two facts suggest that while the conventional wisdom may be appropriate for permanent changes in earnings, it does not explain the reaction of stock prices to transitory changes in economic activity. Investors know that firms probably will not increase dividends by much, if at all, in response to a transitory increase in earnings. They also know that interest rates are likely to rise. But then stock prices must fall in order to reflect the fact that the more or less constant expected dividend stream is being discounted at higher rates. Thus, stock prices should react to news about transitory changes in GDP in much the same way as bond prices.
One part of the puzzle concerns the extent to which investors revise their dividend forecasts in response to news about transitory fluctuations in output and employment. To investigate this issue, I estimated a simple forecasting model for dividends and the cyclical component of private sector GDP. The former are measured by real dividends on the value-weighted New York Stock Exchange portfolio, and the latter is measured by the ratio of private sector GDP to consumption. Economic theory suggests that this ratio is a simple and powerful measure of the cyclical component of GDP (see Cochrane 1994). According to the Permanent Income Hypothesis, most households prefer a smooth path of consumption to a variable one, and they smooth over transitory changes in income by borrowing or lending. Thus, a change in income that is not accompanied by a change in consumption is likely to be transitory. For example, when households experience a temporary decline in income, they borrow to maintain consumption near their customary levels, and they repay their debts later on when income returns to higher levels. In this case, we would see a decline in the income-consumption ratio, and this would correctly signal that households believe that the change in income is transitory.
According to my simple forecasting model, investors make their forecasts each quarter based on current and lagged information about dividends and the income-consumption ratio, and then they wait to see how their forecasts turn out. Inevitably, some unforeseen events occur and cause actual outcomes to differ from the predicted values. When the next quarter comes around, investors learn about these events and observe their forecast errors. Thus, the errors in the forecasting model represent new information about dividends and transitory movements in output. Investors learn from this news and revise their forecasts accordingly. The figure illustrates how investors would revise their dividend forecasts in response to typical bits of news about dividends and cyclical movements in output. Here, “typical” means an average sized forecast error in each of the variables. Evidently, most of the variation in expected dividends results from news about dividends themselves. For example, in response to an average sized forecast error in dividends, investors would revise their dividend forecasts at all horizons upward by about 5 percent. At constant discount rates, this would also raise stock prices by around 5 percent. As one would expect, news that signals a permanent increase in future dividends is likely to raise stock prices.
In contrast, news about transitory movements in GDP has little influence on expected dividends. In response to an average sized forecast error in the income-consumption ratio, expected dividends increase by less than 1 percent, and this response is statistically insignificant. That is, based on this evidence, we cannot rule out the possibility that news about transitory movements in output has no influence on expected dividends.
Why does news about transitory fluctuations in output have so little influence on dividend forecasts? After all, one would think that the earnings of most firms would rise when private sector GDP increases and fall when private sector GDP decreases. But earnings and dividends are not the same, and there is a great deal of evidence going back to Lintner (1956) that firms prefer to smooth their dividend payments over the business cycle. One explanation for dividend smoothing is based on the fact that managers know more about a firm’s prospects than market participants. A change in dividends would communicate this inside information to the market and would have a big effect on a firm’s equity price. Managers are reluctant to change dividends because this signaling channel would contribute to the volatility of the firm’s stock price. Of course, when there is a permanent change in a firm’s prospects, managers must alter their dividend policy. This is evident in the figure, which shows that news about dividends tends to have a permanent effect on expected dividends. But managers can smooth dividends over the business cycle. If they decline to raise dividends during expansions, they can avoid cutting them in recessions. The retained earnings that accumulate during the upswing are paid out as dividends during the downswing.
Since news about transitory movements in output has little effect on expected dividends, it must be the case that the reaction of stock prices is mostly due to changes in the market discount factor. This variable is unobserved, so I cannot bring direct evidence to bear on this question. However, two bits of indirect evidence support my interpretation.
First, as mentioned above, the expected discount rate is approximately equal to the yield on a Treasury bill and tends to rise in response to news of a cyclical expansion. Since expected dividends are roughly constant and expected discount rates are increasing, it follows that stock prices must fall.
There is also some evidence that the market discount factor is enormously volatile. Again, although this variable is unobserved, one can use security market data to estimate a lower bound on its variance (see Hansen and Jagannathan 1991). For example, when I apply a version of their method to returns on 3-month Treasury bills and the value-weighted New York Stock Exchange portfolio, I find that the quarterly standard deviation of the nominal discount factor must be at least 9.4 percent. To put this in perspective, the standard error of quarterly returns on the NYSE portfolio is 4.4 percent. Evidently the discount factor is more than twice as variable as stock returns themselves! A model that implicitly assumes that it is constant is not likely to describe the market very well.
According to conventional wisdom, stock prices should rise in response to news of strong economic growth, yet we observe that stock prices sometimes fall. The conventional wisdom is likely to be correct when the news concerns a permanent change in activity but not when it concerns a transitory change. Firms smooth dividend payments over the business cycle, and investors price these smooth dividend streams using variable discount factors. This results in a fall in stock prices when the market learns of a transitory increase in economic activity.
Cochrane, John H. 1994. “Permanent and Transitory Components of GNP and Stock Prices.” Quarterly Journal of Economics 109, pp. 241-266.
Hansen, Lars Peter, and Ravi Jagannathan. 1991. “Implications of Security Market Data for Models of Dynamic Economies.” Journal of Political Economy 99, pp. 225-262.
Lintner, John. 1956. “The Distribution of Incomes of Corporations among Dividends, Retained Earnings and Taxes.” American Economic Review 46, pp. 97-113.
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120