Regressive Distributed Lag Research Articles

A statistical analysis of the impact of monetary policy on credit union lending

In this paper, a polynomial distributed lag regression model has been employed to determine the responsiveness of credit union consumer installment lending to changes in monetary policy. Estimation of the model leads to the following conclusions: (a) the model provides an adequate statistical representation of credit union reaction to monetary policy, (b) over time, credit union lending behavior has come to react more quickly to changes in monetary policy, and (c) over time, the degree, or strength, of response of lending behaviour to changes in monetary policy has weakend. The model is also applied to commercial bank consumer installment lending with similar results.

The impact of events upon mass political attitudes in Berlin, 1963/64: an application of polynominal distributed lag regression

The article describes a project which sought to develop and apply a novel approach to studying the impact of events upon mass political attitudes by using time series data and the statistical technique of polynominal distributed lag re gression. This approach offers a fruitful analytical perspec tive for historians, for policy science, for research on the dynamics of public opinion and, beyond that, of social and cultural change.

Energy and employment: a time-series analysis of the causal relationship

The relation between total employment and total energy consumption is analyzed using dynamic time series methods. Based on monthly data for the period January 1973–March 1978, there is a unidirectional causality running from energy to employment, without feedback. This suggests that single equation dynamic (distributed lag) regressions of employment on energy can be treated as reflecting behavioral relationships as opposed to solely statistical forecasting relationships. The constructed model indicates that the employment variable reacts to a change in the energy availability inversely, after a pure delay of eight months. 1 1 Including the month in which the change occurs The response decays geometrically, and it is practically complete by the end of the twelfth month following the shock in the energy variable. Thus, the causal dynamics between the two variables could not have been captured using annual data. The long run elasticity of total employment with respect to domestic energy consumption is estimated to be −0.1356. An increase in total employment should result from a constraint on energy use, rather than the decrease widely predicted.

The St. Louis Equation and Reverse Causation: The Evidence Reexamined

A major criticism of the so-called St. Louis equation, Andersen and Jordan [2], has been that it is not a true reduced form equation since the monetary policy variable is not statistically exogenous with respect to GNP [4; 6; 7; 9], and therefore, it is inappropriate to estimate by ordinary least squares the distributed lag regression of nominal income on monetary and fiscal policy variables. Quite recently, however, statistical techniques1 have been developed to test whether or not the explanatory variables in a regression equation can be regarded as strictly exogenous [16; 17; 18; 15]. Of great interest is the work by Sims [16] who has shown that in a distributed lag regression of nominal income on monetary policy variables, monetary policy variables such as the monetary base can be regarded as strictly exogenous with respect to nominal GNP.

Distributed lag regression by an almost periodic design matrix

Frequency domain estimation and tests of hypotheses are considered for a general multivariate distributed lag regression model. The usual vector of regression functions is replaced by a matrix of almost periodic functions, a case for which the terms appearing in the frequency domain estimators have well-defined limits. Asymptotic distributions and consistency are established for the regression coefficients and error spectra. A test statistic is proposed for the no regression hypothesis which is asymptotically central under the null hypothesis and converges to infinity under the alternative. In the multivariate case, a product of beta-distributed variables is obtained.

Distributed lag regression by an almost periodic design matrix

Frequency domain estimation and tests of hypotheses are considered for a general multivariate distributed lag regression model. The usual vector of regression functions is replaced by a matrix of almost periodic functions, a case for which the terms appearing in the frequency domain estimators have well-defined limits. Asymptotic distributions and consistency are established for the regression coefficients and error spectra. A test statistic is proposed for the no regression hypothesis which is asymptotically central under the null hypothesis and converges to infinity under the alternative. In the multivariate case, a product of beta-distributed variables is obtained.

An Empirical Note on Some Monetarist Propositions

The recent conference on monetarism underlined, among others, two important propositions advanced by monetarists that are a subject of controversy both at the theoretical and the empirical level. These propositions are: (1) changes in the growth of nominal income and the general price level and short run changes (relative to trend) in the real income are primarily the results of changes in the growth rate of the money supply and (2) the long run movements in real economic variables, such as real income, are little influenced, if at all, by monetary actions [15, 1-6]. The main purpose of this note is to present empirical evidence which is consistent with the above monetarist propositions. It is shown that the estimation of the simple bivariate distributed lag relations between income (nominal as well as its real and price components) and the monetary policy variable yield inferences which are consistent with the above two propositions. Central to this inquiry is the use of a distributed lag estimation methodology in which the length of the lag distribution is kept very generous to study the long term properties of the monetary effects and the use of tests of statistical exogeneity recently proposed in Sims [14]. The plan of this note is as follows. Section I presents the estimates of the simple bivariate distributed lag regressions for income and its components and discusses the associated statistical inferences. Section II then talks about some econometric objections that can be labeled against such estimated equations and presents some alternative tests results consistent with the monetarist propositions. Section III contains the summary and concluding remarks.

Is Money Exogenous in Money-Demand Equations

Sims's finding that nominal money stock is strictly exogenous in a distributed-lag regression of nominal income on nominal money stock is not inconsistent with the appearance and real income and nominal interest rates as strictly exogenous regressors in the quarterly money-demand equations estimated in real form. This strict exogeneity of real income and interest rates in real money-demand equations is due to the restriction implied by estimation in the real form, and these implicit restrictions seem to conflict with the sample information.

Analysis of Employment in the Manufacturing Sectors of Nine Urban Areas: The Relation between Average Hours Worked and Employment

This paper investigates the short-term relationship between changes in the average number of hours worked and the number of workers employed. Nine medium-sized urban areas in Michigan have been chosen for the study, which deals with production workers engaged in manufacturing. It is assumed that changes in external demand for manufactured goods directly effect the amount of labor used by the manufacturing sector of the nine areas. An increase in demand causes an increase in labor, while a decrease in demand has a negative effect on labor. Firms may adjust either the number of hours worked by their production workers or the number of workers they employ, or both. Correlation analysis is used to investigate the existence of leads or lags in the relationship between the two variables. A distributed-lag regression function is employed to examine the quantitative importance of this relationship. The results of correlation suggest a tendency for the two aggregate variables to change simultaneously or with short time lags. The time pattern of the relationship depends on a number of factors, the most important of which are discussed in the paper. The results of regression indicate that the strength of the quantitative relationship can vary from area to area when the relationship between employment and hours worked during the same period is considered, while no significant regional differences are observed in the strength of the relationship between employment and hours worked in previous periods. Some policy implications of the results are discussed at the end of the paper.

Short-term relationship between manufacturing and non-manufacturing employment in nine urban areas

This paper explores the short-term relationship between manufacturing and non-manufacturing employment in nine medium-sized urban areas in Michigan. More specifically, it examines the extent to which changes in employment in the non-manufacturing sector seem to be associated with changes in manufacturing employment; the existence of possible leads or lags in the relationship of the two employment sectors; and the quantitative importance of this relationship. It is assumed that, since the manufacturing sectors in the above areas export the largest part of their production, the fluctuations which occur in the demand for manufactures and in manufacturing employment are mainly due to external factors. The demand for non-manufacturing products and employment is then expected to fluctuate primarily as a result of the changes in the manufacturing sector. Correlation analysis is used to examine if there is any association in the variation of the two sectors and to detect any leads or lags. The results of this analysis suggest strong simultaneity in the variation of the two variables which may reflect the existence of unemployment and underemployment in certain industries and the influence of fluctuations that occur in the construction sector. A KOYCK-type distributed-lag regression function is used to investigate the quantitative importance of the relationship. The results of regression analysis suggest that a change in the number of manufacturing workers may cause a change in the number of non-manufacturing workers but the strength of the effect may vary according to the urban area. Some reasons for this are suggested in the paper.

Consumer Characteristics Associated with Dealing: An Empirical Example

The targeting of consumer deals, as well as our understanding of consumer behavior, would be enhanced by knowledge of what consumer characteristics relate systematically to deal purchasing activities in a product market. This study examines possible relationships between a housewife's dealing activity in a product class (dentifrice) and certain of her social-psychological and purchasing characteristics. Nearly one billion dollars was spent for coupons distributed by direct mail, magazines, and newspapers, in 1960 [5]. To this must be added the cost of price-off labels, premium packs, and other consumer deals. Yet in spite of its economic significance, there has been little study of market response to dealing. Notable exceptions are [2, 3, 6]. Two previous studies of dealing are relevant to this one. Webster [6] analyzed the relationship between a family's deal proneness index and 11 socioeconomic and 34 purchasing characteristics. The deal proneness index reflected a family's propensity to deal more or less than expected in a product class. He found brand loyalty (proportion of purchases concentrated on the favorite brand) to be negatively related to deal proneness. Of the socioeconomic variables, only age of housewife related significantly to deal proneness. Similarly, Massy and Frank [3], in distributed lag regressions of weekly price and dealing on a brand's market share, found that brand loyal consumers were less responsive to dealing than were nonloyal consumers. The picture that emerges from the limited empirical work to date is that brand loyalty and dealing tend to be inversely related and that most traditional socioeconomic measures are not useful predictors of dealing activity. THE VARIABLES

Almost-Periodic Signal Estimation

Previous article Next article Almost-Periodic Signal EstimationA. S. KholevoA. S. Kholevohttps://doi.org/10.1137/1116024PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAbout[1] Chiang Tse-Pei, On the estimation of regression coefficients of a continuous parameter time series with a stationary residual, Theory Prob. Applications, 4 (1959), 373–390 10.1137/1104037 LinkGoogle Scholar[2] Yu. A. Rozanov, Gaussian infinite-dimensional distributions, Trudy of the V. A. Steklov Mathematical Institute, 108 (1968), , (In Russian.) Google Scholar[3] N. K. Bari, Biorthogonal systems and bases in Hilbert space, Moskov. Gos. Univ. Učenye Zapiski Matematika, 148(4) (1951), 69–107, (In Russian.) MR0050171 Google Scholar[4] V. V. Stepanov, On a class of almost periodic functions, Doklady Akad. Nauk SSSR (N.S.), 64 (1949), 297–300, (In Russian.) MR0028453 Google Scholar[5] V. F. Pisarenko and , Ju. A. Rozanov, On certain problems for stationary processes leading to integral equations related to Wiener-Hopf equations, Problemy Peredači Informacii No., 14 (1963), 113–135, (In Russian.) MR0192564 Google Scholar Previous article Next article FiguresRelatedReferencesCited byDetails Introduction to Random Processes, With Applications to Signals and SystemsTechnometrics, Vol. 29, No. 2 Cross Ref Regression for randomly sampled spatial series: the trigonometric case14 July 2016 | Journal of Applied Probability, Vol. 23, No. A Cross Ref Distributed lag regression by an almost periodic design matrix14 July 2016 | Journal of Applied Probability, Vol. 15, No. 4 Cross Ref Volume 16, Issue 2| 1971Theory of Probability & Its Applications History Submitted:28 October 1968Published online:17 July 2006 InformationCopyright © Society for Industrial and Applied MathematicsPDF Download Article & Publication DataArticle DOI:10.1137/1116024Article page range:pp. 249-263ISSN (print):0040-585XISSN (online):1095-7219Publisher:Society for Industrial and Applied Mathematics

The Effect of Advertising on Liquor Brand Sales

This study estimates the revenue and profit generated by increments of advertising expenditures for liquor brands. Various distributed-lag regressions were fitted to ten-year periods for 15 brands ranging from small to largest. Found were annual retention rates around .75, and a logarithmic response function. Perhaps most important, this technique, pioneered by Telser and Palda, is again found useful.