Knowledge that Transforms

Ernest H. Oksanen, James R. Williams, A Principal Components Analysis of the National Accounting Data of Eighteen OECD Countries, The Review of Economics and Statistics, Vol. 56, No. 4 (Nov., 1974), pp. 559-562

EXTRAORDINARILY high unemployment among nonwhite teenagers is often explained in terms of the disadvantaged environment from which this group comes. As noted in a recent Manpower Report of the President (1972, p. 80), however, there is a large gap between whites and blacks in the extent of unemployment within urban poverty areas. This suggests that even among males living in the inner-city, racial discrimination in employment may be important in explaining observed racial differentials in unemployment rates. Isolation of the direct impact of race on unemployment is difficult because variations in many factors related to unemployment for both whites and nonwhites are also associated with race as a consequence of generations of discrimination in housing, education, employment, and participation in social and political processes. This study attempts to measure the direct impacts of race and age in determining racial differentials in unemployment rates among males aged 16-21 and 22-34 years residing in urban low-income areas. The older group of males is examined because while the unemployment rates of both whites and blacks drop sharply as teenagers become young adults, the unemployment rate of blacks in recent years has fallen further than that of whites resulting in a lower black-white ratio 'of unemployment rates for the older age category (see Leigh and Rawlins, 1973). Two specific issues are addressed in the paper. (1) How much of racial unemployment rate differentials can be explained by race after standardization by employment-related personal characteristics, and how does the magnitude of the differentials explained by race differ across age groups? (2) How much of the lower unemployment rates observed for the older age group can be attributed to age after standardization, and does the effect of aging differ by race? In approaching these questions we utilize a new source of data, the Census Employment Survey (CES) (United States Bureau of the Census, 1972), recently made available as a part of the 1970 Census. The survey provides data from persons living in 60 selected lowincome areas in 51 large cities and 7 rural areas.1 Approximately half the respondents are black and nearly 12 % are Spanish-speaking (largely Puerto Ricans, Chicanos, and Cubans). For the purposes of this study, CES data offer the advantage, relative to Census data, of providing more information on job training, job-seeking methods, job tenure, and other cognitive skills usually thought to be associated with success in the labor market.

IN his Economics and Information Theory, Theil (1967) developed a theorem which states that under certain conditions the covariance between the variation in the demand for two goods, holding constant prices and income, would measure the substitutability of the two goods (more precisely, will measure their Slutsky-Hicks substitution effect up to a negative scaler) (chapter 7). This theorem and the framework behind it have subsequently been used in further studies of commodity preferences and substitution (Barten (1968), Phlips (1971), and Phlips and Rouzier (1972)). If the covariance measure of substitution could be usefully applied to asset substitution, it could prove to be a valuable supplement to the more traditional measure of asset substitution, i.e., the estimation of relations between asset demand and asset yields. Theoretically, the measure can be applied to time-series or crosssection data. The measure might have a wider applicability to cross-section data than the traditional asset demand-yield measure since the latter can only be applied to assets with regional markets (e.g., deposits). Moreover, to some extent, it can be expected that errors or biases that might arise in the covariance measure would be independent of those incurred in using the traditional measure of asset substitutability. This paper will attempt to apply the covariance measure to estimating asset substitution among households. Part I considers the theoretical underpinnings of the covariance measure and its theoretical relation to the asset demandasset yield measure of substitutability. In part II an empirical formulation of the covariance measure is developed and applied to cross-section asset data for households and the results analyzed. The results are then compared to two earlier, and more limited, asset studies which used essentially a covariance type measure to assess substitutability. The results are also compared to those of household studies which have used the traditional method for estimating substitution among assets. Findings and suggestions for further study are summed up in the conclusion.

THE purpose of this paper is (1) to determine whether corporations' decisions to borrow in the public or private corporate bond markets are significantly influenced by yield differentials between the two markets and (2) whether the sensitivity of corporations to yield differentials increases with the relative ease with which they can borrow in either market. In recent years, the influence of interest rates on corporation financing decisions has received considerable attention in studies which have sought to identify the determinants of corporate external financing and the maturity composition of corporations' liabilities.' In none of these studies, however, can the signs of the interest rate variables in the regression equations be predicted. This is because the expectations hypothesis of the term structure of interest rates which these studies accept posits that interest costs will be equalized regardless of the maturity financing strategy a borrower adopts. As a result the credibility of the findings of these studies concerning the influence of interest rates on financing decisions must be questioned even when this influence was found to be significant.2 In contrast, yield differentials between the public and private corporate bond markets have a predictable effect on the distribution decisions of borrowers. Since nominal rates typically are higher in the private market, a decrease (increase) in yield differentials should encourage (discourage) the sale of private placements. In the aggregate, then, the percentage of corporate debt sold privately should be negatively related to yield differentials between the two markets. In section II, the variables which influence the supply of and demand for funds in the public and private markets -are identified, and supply and demand equations are formulated. Differences in the distribution process in the two markets as well as marketing constraints, limit the ability of some corporations to borrow in both markets, as described in section III. For the purpose of measuring the influence of yield differentials on distribution decisions, then, corporations are divided into three groups, based on their ability to borrow in both markets. The regression results are discussed in section IV. Only the demand equation is estimated since appropriate data on the supply of funds to the corporate market is not available. The estimation of demand by itself, however, produces biased estimates of the coefficient of yield differentials between the public and private markets. The direction of this bias is considered in section V. The major findings are that yield differentials have exerted a significant influence on the distribution decisions of most of the groups studied. Moreover, the sensitivity of corporations to yield differentials is an increasing function of the relative ease with which they can borrow in both the public and private markets. Finally, if supply and demand had been estimated simultaneously, yield differentials most likely would have exerted a stronger influence on the distribution decisions of borrowers. Received for publication November 29, 1972. Revision accepted for publication May 22, 1973. * This paper is based in part on a study of the private placement market by Shapiro and Wolf (1972). In Shapiro and Wolf (chapter 5), the variables influencing the supply and demand for public and private financing are discussed and the distribution patterns of different borrower categories are compared. This paper extends this analysis by estimating a demand equation for public and private financing and by measuring the sensitivity of different borrower groups to yield differentials between the two markets. T am indebted to Professors John Keith, Richard Rippe and Maurice Wilkinson of the Columbia Business School and to Professors Robert Glauber and John Lintner of the Harvard Business School for their helpful comments. This study was supported by faculty research funds of the Columbia Business School. 1 Cragg and Baxter (1970) attempted to determine the factors which influenced the timing and the size of bond financings of 129 industrial companies using a probit analysis. Bosworth (1971) and White (unpublished) explained the maturity composition of total corporate borrowing using the Federal Reserve's Flow-of-Funds data. 2 Only in White's study was the fraction of long-term-tototal debt issues significantly (and negatively) related to the relative cost of long-term financing.

r HE increased attention in the literature devoted to the market for federal agency securities,' parallels the rapid growth in that sector of the capital market since 1966. Two studies, one by Kochin (1972)2 and the other by Peskin (1971),3 have already appeared. These studies surveyed various aspects of the agency market, such as the structure of yields in the secondary market, the characteristics of agency securities, and certain indicators of performance in the secondary market, such as volume traded, bid-ask spreads, dealer positions and price variability. This paper is addressed to a more specific aspect of the agency market, namely, the determinants of the yield on newly-issued agency securities. In order to avoid the issue of determining the overall level of interest rates we reduce the problem to determining the yield spread between newly-issued agency securities and comparable maturity Treasury securities. A comparable maturity Treasury issue is considered the best available benchmark with which to compare newly-issued agency issues.4 A key issue which receives considerable attention below is whether there exists an optimum size of issue in the agency market. In other words, we will try to isolate the size of issue which minimizes the yield spread between an agency issue and a comparable maturity Treasury security. Minimizing this yield spread is an important objective for a federal agency since this is one criterion used to judge the efficiency with which it issues its liabilities in the capital market.5

JHE PURPOSE of this article is to bring to the attention of the readers of this journal a number of related and important estimators that are currently being discussed in the statistical literature which have implications for applied work since rules are employed which seek to improve the performance of conventional estimators. In spite of the rapid advances, over the last three decades, of economic theory, econometric procedures, and data relating to economic processes and institutions, the search for quantitative economic knowledge still remains to some extent an essay in persuasion. In the process of nonexperimental model building there are typically many admissible economic and statistical models which do not contradict our perceived knowledge of human behavior. Thus, in model specification there is usually uncertainty, for example, relative to the algebraic form, classification, number and timing of variables to be included in the behavioral and technical relations, and the corresponding stochastic assumptions. When econometric models are correctly specified, statistical theory provides procedures for obtaining point and interval estimates and evaluating the performance of various linear and usually unbiased (at least asymptotically) estimators. But, the applied worker must inevitably work with false models, where the true specification of the sampling model is unknown. Furthermore, the statistical model employed is usually determined by some preliminary testing of hypotheses using the data at hand. This search procedure, involving two-stage or repeated significance test procedures applied to the same set of data and yielding an estimate after the preliminary test(s) if significance, is often used in applied work in economics with little or no information on the sampling properties of the resulting estimator and with little or no consideration to the possible distortion of subsequent inferences. Within this context, we, seek to generalize and extend the results of Wallace and Ashar (1972) relative to preliminary test or two-stage estimating procedures and call attention to another important class of estimators and estimator comparisons. In particular, we review the possible statistical consequences of using preliminary test or sequential estimators in the search process and suggest old and new estimators, that are superior, under a squared error loss measure for gauging estimator performance, to the conventional estimators usually employed. We also note that conventional estimating procedures currently used in applied work may not be appropriate for the problem at hand and, perhaps more importantly for the researcher, we show that better alternative estimators exist. Perhaps it is appropriate at this point to note that, when making a choice between estimators, the traditional solution is to restrict consideration to the class of linear unbiased estimators and hope that among the estimators in the restricted class, one has uniformly smallest risk. Fortunately for many problems a best linear unbiased estimate exists. In this paper, in discussing the estimators that are alternatives to the conventional least squares estimator, we will leave the class of linear unbiased estimators. The notion of unbiasedness which has been accepted by or perhaps forced on applied workers, although intuitively plausible, is an arbitrary restriction or property and has no direct connection with the loss due to incorrect decisions. The economist who is interested in parameter estimates or predictions appropriate for choice purposes, may not care if he is right on the average, and thus the unbiasedness property may be unsatisfactory from a decision point of view. In any event our purpose, which is to some extent expository in nature, is to focus on point estimation under a squared error loss measure of goodness and bring the statistical consequences of making use of conventional and Received for publication October 2, 1972. Revision accepted for publication April 3, 1973. *Arnold Zellner read an early draft of this paper and made many helpful comments.