Working Paper Abstracts – 1992
Many recommendations for reforming securities market are predicated on the belief that providing information on order flow and other market variables to traders (i.e., increasing market transparency) will increase liquidity and improve price efficiency. This paper demonstrates that market transparency can actually increase price volatility and lower market liquidity. This occurs even though transparency increases the precision of traders’ predictions about the asset’s value. In a sufficiently large market, transparency always reduces volatility and improves market quality. We use these results to assess policy proposals concerning the disclosure of trading information.
Traditional theories of asset pricing assume there is a complete market participation, in the sense that all investors participate in all markets. In that case, preferences shocks typically have only a small effect on asset prices and are not an important determinant of asset price volatility. However, there is considerable empirical evidence that most investors participate in a limited number of markets. We show that limited market participation can amplify the effect of preference shocks, so that an arbitrarily small degree of aggregate uncertainty about preferences causes a large degree of price volatility. We also show there may exist Pareto-ranked equilibria, where the Pareto-preferred equilibrium is characterized by a different pattern of participation and lower volatility.
This paper reexamines evidence on the monotonicity of the term premium. Using a recently developed approach for testing inequality constraints, we propose and conduct tests for whether the term premium is monotonic and reach different conclusions from those implied by individual t-statistics on term premiums, even under a Bonferroni-type adjustment. Our results generally support McCulloch’s (1987) view that the liquidity preference hypothesis remains unrefuted.
The purpose of this paper is to compare execution prices of NYSE-listed stocks on the NYSE and on non-NYSE markets. The first conclusion of this comparison is that most of the time the NYSE had the best quote. This result does not necessarily imply that execution prices on the NYSE are better than the regionals since an investor should always receive an execution price no worse than the best intermarket quote regardless of the particular market on which an order is executed. The second conclusion is that there is more price improvement on the NYSE than on other markets for NYSE-listed stocks. The third conclusion is that the average price improvement from trading on the NYSE varies according to the price of the stock and the size of the transaction. On a 100-share transaction of a 40-dollar stock, the price improvement on the NYSE is on average 1.5 cents greater than that on non-NYSE markets.
Targeting of broadly based monetary aggregates, such as M2, continues to play an important role in U.S. monetary policy. Yet the average reserve requirement on M2 assets has fallen dramatically in recent years, a trend which may hamper the ability of the central bank to control such large aggregates. This paper explores the controllability and information content of monetary aggregates in a simple, flexible-price macroeconomic model, with specific attention to the role played by reserve ratios.
It is found that although reserve ratios influence the conditional expectations of both the price level and real output, they do not influence the variance of these variables conditioned on observations of nominal monetary aggregates. The best indicator of real income is a deposit aggregate whose income elasticities differ the most from the income elasticity of currency while, in general, a deposit aggregate with a very low income elasticity is the best indicator of the aggregate price level. These results suggest that broad monetary aggregates such as M2 may be better indicators of the price level while narrow aggregates such as M1 may be better indicators of real income. Finally it is shown that the factors which make deposits more controllable by the central bank are often incompatible with the factors that make these aggregates good indicators of real income and prices.
This manuscript reviews the evolving literature on the pricing of assets and the structure of financial markets. It begins with the development of early stock valuation models and proceeds to a description of the efficient-market hypothesis. The survey then examines recent empirical data that has led to a reevaluation of the assumptions underlying an efficient market and ends with a brief survey of the recent literature on market-making and how these models relate to studies of efficient markets.
Recent work in macroeconomics has suggested that (S,s) rules are a useful characterization of household purchases of durable goods, and that they provide a basis for expenditure dynamics consistent with those observed in aggregate data. The purpose of this paper is to determine whether a flexible (S,s) rule, that is, an inaction range that varies over time, provides a more complete characterization of the aggregate data. The results suggest that not only does a flexible rule bring additional explanatory power, but the manner in which the rule implicitly changes over time is highly cyclical and correlated with changes in financial volatility over the business cycle. These results provide an economic structure for the observations that durables purchases contract when consumer uncertainty increases.
We present a test of the theory of the term structure developed by Cox, Ingersoll, and Ross (CIR). The econometric method uses Hansen’s Generalized Method of Moments and exploits the probability distribution of the single state variable that determines real bond prices. The approach avoids problems due to measurement errors in bond prices; it does not employ data on aggregate consumption; and it enables the estimation of a continuous time model based on discretely-sampled data. The tests indicate that the model for real indexed bonds that underlies all the alternative specifications in CIR performs reasonably well when confronted with short-term Treasury Bill data. The parameter estimates indicate that term premiums are positive and that the term structure of indexed bonds can admit several shapes. However, we find it difficult to rationalize the sample serial correlation in Treasury Bill returns using our estimates of the CIR parameters.
This paper derives simple closed-form solutions for expected rates of return on stocks and riskless one-period bills under the assumption that shocks to the growth rates of consumption and dividends are generated by a Markov regime-switching process. These closed-form solutions are used to show that the Markov regime-switching process exacerbates the equity premium puzzle and the risk-free rate puzzle. Three empirical examples illustrate the magnitude of the effects of Markov regime switching on equilibrium expected returns.
The methods of Gibbons and Ferson (1985) are extended, relaxing the assumption that expected returns are linear functions of predetermined instruments. A model of conditional mean variance spanning generalizes Huberman and Kandel (1987). The empirical results indicate that more than a single risk premium is needed to model expected stock and bond returns, but the number of common factors in the expected returns is small. However, when size-biased common stock portfolios proxy for the risk factors, we reject the hypothesis that four of them describe the conditional expected returns of the other assets.
This paper analyzes the risks and returns of different types of real estate-related firms traded on the New York and American stock exchanges (NYSE and AMEX). We examine the relation between real estate stock portfolio returns and returns on a standard appraisal-based index, and find that lagged values of traded real estate portfolio returns can predict returns on the appraisal-based index after controlling for persistence in the appraisal series. The stock market reflects information about real estate markets that is later imbedded in infrequent property appraisals. Additional analysis suggests that the differences in the return and risk characteristics across different types of traded real estate firms can be explained in part by appealing to real estate market fundamentals relating to the degree of dependence of the real estate firm upon rental cash flows from existing buildings. These findings highlight the heterogeneity of securitized real estate-related firms.
Large investors frequently receive some type of research service in return for sending an equity trade to a specific brokerage firm. The commissions so directed are associated with the term “soft dollars.” This study employs a survey of institutional managers and a sample of their trading records to examine the effect of soft dollars on the structure of the brokerage industry and finds that these effects have been significant in redirecting order flow to different types of brokers. Additionally, there is some evidence that investment managers tend to send their easier orders to brokerage houses providing research for soft dollars and their harder orders to more traditional brokerage houses that are more likely to commit their own capital to facilitate a trade. Furthermore, investment managers tend to be less pleased with the quality of execution for certain types of soft dollar transactions.
Recent empirical work has uncovered U-shaped patterns of large magnitude in the serial correlation estimates of multi-year stock returns. The current literature in finance has taken this evidence to mean that there exists a temporary component of stock prices. This paper provides an alternative explanation regarding these findings. Specifically, we show that the patterns in serial correlation estimates and their magnitude observed in previous studies should be expected under the null hypothesis of serial independence.
It is a common empirical finding that stocks are not a good hedge for either ex-ante or ex-post inflation at short horizons. In contrast, using two centuries of data, we demonstrate that there is a positive relation between stock returns and inflation at long horizons. This result is reconciled with the anomalous short run evidence by appealing to the proxy-effect in the context of a consumption based asset pricing model.
In some countries, such as the U.S. and the U.K., stock markets have played an important role in the allocation of resources while in others, such as France, Germany and Japan, only banks have played an important role. This paper investigates the relative advantages that stock markets and banks have in providing finance for industry. It is argued that stock markets are desirable when investors disagree on the optimal policies a firm should pursue because they allow diversity of opinion to be taken account of; this type of situation is likely in new industries or those with an oligopolistic or monopolistic structure. Banks, on the other hand, are desirable when there is wide agreement on optimal policies as in most competitive industries.
We present a finite period general equilibrium model of an exchange economy with asymmetric information. We say that a rational expectations equilibrium exhibits an expected bubble if the price of an asset in one period is higher than any agent’s marginal valuation of holding the asset to maturity. We say the equilibrium exhibits a strong bubble if the price is higher than the dividend with probability one. We show that a necessary condition for an expected bubble to exist is that each agent must be short sale constrained at some period in the future with positive probability. We show that necessary conditions for a strong bubble to occur are that (1) each agent must have private information in the period and state in which the bubble occurs and (2) agents’ trades are not common knowledge. We also present examples of rational expectations equilibria that exhibit strict bubbles when the necessary conditions are satisfied.
This paper proposes and conducts direct tests of the mixture of distributions model for stock prices. By exploiting the model’s bivariate conditional normality of price changes and trading volume, these restrictions can be tested under very weak assumptions regarding the daily flow of information to the market. As a technical byproduct, important parameters governing the distribution of this unobservable information flow are estimated.
Previous research has investigated the multivariate normality of stock returns using tests based on the marginal distribution of returns. Due to the contemporaneous correlation across asset returns, these tests are difficult to interpret. We develop a general test procedure which takes account of the correlation across assets and which focuses on both the marginal and joint distributions of returns. We find highly significant evidence that stock returns and market model residuals are nonnormal. Moreover, this nonnormality appears in both the marginal and joint distributions of asset returns.
The incorporation of diverse information into asset prices is empirically examined in an actual securities market with multiple rounds of trade. Using prices of Israeli index and nominal bonds of equal maturity, we calculate implied expectations of inflation that has already occurred but for which the official statistic has not yet been announced. ‘Learning’ is defined as the convergence of these expectations to the actual level of inflation in the period after the end of the month but before the announcement of the official statistic. We find that the variance of the inflation expectation errors decreases with trading days in this period. The decline in the variance suggests that investors learn, by repeatedly observing prices, about the distribution of other investors’ information. We also find a positive relation between the dispersion of relative price changes and the size of the inflation-expectation errors on the first round of trade. The correlation diminishes as investors learn about the distribution of inflation information in the economy.
This paper examines the role of market makers in facilitating price discovery. We show that a specialist may experiment with prices to induce more informative order flow, thereby expediting price discovery. Market makers in a multiple dealer system, unlike a specialist system, do not have the incentives to perform such costly experiments because of free-rider problems. Consequently, the specialist system may provide open markets where competition fails, but at the cost of wider bid-ask spreads. We analyze the effect of experimentation on the bid-ask spread and provide an exploratory analysis of intraday specialist data which is consistent with our price experimentation hypothesis
This paper examines the effects of large (block) transactions on the prices of common stocks. We develop an explicit model of the mechanism by which block trades are accomplished. The model yields several empirical hypotheses that are tested with a unique new data set. The data set covers nearly 4,240 transactions in the period 1985-1990, primarily in less liquid stocks. The blocks can be identified as either seller-initiated or buyer-initiated, and are all arranged in the ‘upstairs’ off-exchange market by either exchange members or non-exchange members. This knowledge is critical since the direction of hypothesized price effects depend on who initiates the trade and where it is executed. We find price impacts are positively related to trade size, and that this relation is non-linear, as predicted by the model. Trades executed on the OTC market have significantly larger price impacts than the impacts for comparable trades on the NYSE or AMEX. Further, there are significant differences in the price response to buyer-initiated and seller-initiated block trades.
This paper develops a model of market making that incorporates both inventory control and asymmetric information effects. We show that the specialist acts both as a market maker and as an active investor trading for his own account. As a market maker, the specialist quotes prices that induce mean reversion toward a desired level of inventory; as an active investor, he periodically adjusts the target inventory levels towards which inventories revert. We test the model using data obtained from a NYSE specialist. We find that specialist inventories exhibit mean reversion, but the adjustment process is slow, even controlling for shifts in target inventories. The model also predicts that quote revisions are negatively related to specialist trades and positively related to the information conveyed by order imbalances. We find strong evidence for this hypothesis; further, our results suggest that specialist quotes anticipate future order imbalances.
Recent empirical findings suggest that equity returns are predictable. These findings document persistent cross-sectional and time series patterns in returns that are not predicted by extant theory, and are, therefore, often classified as anomalies. In this paper we synthesize the evidence on predictable returns, focusing on the subset of the findings whose existence has proved most robust with respect to both time and the number of stock markets in which they have been observed.
Are stock prices determined by fundamentals or can “bubbles” exist? An important issue in this debate concerns the circumstances in which deviations from fundamentals are consistent with rational behavior. When there is asymmetric information between investors and portfolio managers, portfolio managers have an incentive to churn; their trades are not motivated by changes in information, liquidity needs or risk sharing but rather by a desire to profit at the expense of the investors that hire them. As a result, assets can trade at prices which do not reflect their fundamentals and bubbles can exist.
We estimate a multiproduct cost function model that incorporates measures for the quality of bank output and the probability of failure, which can influence a bank’s costs in a variety of ways. We model a bank’s price of uninsured deposits as an endogenous variable depending on the bank’s output level, output quality, financial capital level, and risk measures. Incorporating these aspects into the cost function has a significant effect on measures scale and scope economies when compared with results of previous studies that did not take quality and risk into account. We find constant returns to scale at the mean-sized bank and at banks in four different size categories. We also find evidence of diseconomies of scope at the larger banks. Finally, there is evidence that the “too-big-to fail” doctrine has a significant impact on the price a bank pays for its uninsured deposits. For banks in the largest size category, an increase in size, holding default risk and asset quality constant, significantly lowers the uninsured deposit price.
I modify the stochastic econometric cost frontier approach to investigate efficiency in mutual and stock S&L using 1991 data on U.S. S&Ls. My methodology allows both the cost frontier and error structures to differ between S&Ls of these two ownership forms. A likelihood ratio test indicates that the data support this unrestricted model, which implies efficient mutual and stock S&Ls use different production technologies. Various measures of inefficiency show that on average stock S&Ls are less efficient than mutual S&Ls. The second part of the article relates the inefficiency measures to several correlates.
This study explores the integration of the markets for NYSE-listed stocks. Although the NYSE bid or offer is part of the best displayed intermarket quote roughly ninety percent of the time, there is some evidence that non-NYSE markets do on occasion contribute to price discovery. Actual execution prices for NYSE-listed stocks sometimes fall within the best displayed intermarket quote. The average effective spread across all stock is 13.5 cents, which is 76.2 percent of the best displayed intermarket spread. Often, the best displayed spread is almost double the average effective spread. For 100- and 200-share prints, the average effective spread on the NYSE is consistently less than on non-NYSE markets; but for prints of 1,000 to 3,000 shares, the reverse sometimes occurs.
This paper develops tests of inequality restrictions implied by conditional asset pricing models. The methodology is easy to implement, requires little knowledge of the conditional distribution of asset returns, and is valid under fairly weak assumption. We provide several examples of asset pricing models in which inequality constraints play a central role, documenting results which are in contrast to recent empirical work in these areas. Specifically, we document reliable evidence that (i) the ex-ante risk premium is not always positive, (ii) the size effect does not hold conditionally, and (iii) term premiums may be monotonically increasing in maturity.