THE DEVELOPMENT OF NATIONAL ACCOUNTS IN ARGENTINA*

The history of national accounting in Argentina is presented in brief. The use of the production method as a basis for GDP estimates is explained and sources and methods of deriving the estimates for different sectors are commented on in some detail. Next the reliability of the estimates is examined for sectoral product, national income by factor shares and the components of final expenditure in terms of the comparison of two different estimates for the period 1950–1963. The degree of accuracy is judged to be generally sufficient, but the importance of economic censuses in the process of estimation is stressed. The importance of detailed studies on several different aspects of the economic structure (input-output, personal and family income by size and regional accounts) is also stressed as a basis for improving the reliability of estimates.     

Accuracy and reliability of Spanish regional accounts (CRE-95)

Are estimates accurate? Are these data reliable? Can we evaluate whether successive revisions of the first estimate bring additional information, or whether they are due to the correction of earlier inaccuracies of bias, which does not add any useful information to economic analysts? The response is that a statistical technique, known in economic literature as news versus noise analysis can be used to answer these questions. These concepts were first introduced in ground breaking articles by Mankiw et?al. (1984) on revisions to money stock and Mankiw and Shapiro (1986) on revisions to Gross National Product in the United States. This method claims that the proximity of the first estimate to the definitive one can be a clue as to its quality. Many studies investigate the magnitude and nature of errors contained throughout the National Account System estimates; however, economic literature offers no option for judging accuracy and reliability in the main Regional Account aggregates, and therefore, this methodology is adopted to assess the accuracy and reliability of the revision process of the Gross Value Added (GVA) growth rates at basic prices, 1995 Base, within the Spanish Regional Accounts (CRE-95). For that purpose, this study examines five comprehensive revisions analyses for a range of eighteen Autonomous Communities (NUTS-2) during 1999. Main conclusions are: the quality of the estimates tends to increase over time; and the process can be explained better as reflecting new information than as correcting measurement errors so that the National Statistics Institute??s procedure can be deemed to be suitable.     

ASSESSING THE VALUE OF ECONOMICS RESEARCH: THE CASE OF THE BIAS IN THE CONSUMER PRICE INDEX

Economists provide sometimes contradictory information about economic systems that contribute to policy design. How does one value this type of knowledge? A political‐economic game is presented that allows for reinforcing and contradictory research messages. Policy makers are assumed to follow a Bayesian decision theory process and the model is tested with quantitative estimates of the value of research on the degree of bias in the Consumer Price Index. Most economists agree that a bias exists, but published estimates vary widely. A blue‐ribbon panel of economists recommended revisions to how the index is calculated, and the Bureau of Labor Statistics subsequently made revisions that differed from their original plans, but how much influence did the panel really have on the revisions? (JEL Z00)     

THE CHOICE OF THE AGGREGATION LEVEL IN THE ESTIMATION OF QUARTERLY NATIONAL ACCOUNTS

Many central statistical offices use indirect time series disaggregation methods to produce quarterly national accounts estimates or other high frequency variables. This paper investigates the relation existing between the statistical properties of indirectly estimated time series and the contemporaneous aggregation level at which estimation is carried out, when a version of the Chow-Lin (1971, 1976) method is used to evaluate quarterly time series. It is shown that estimation at the lowest possible level of contemporaneous aggregation is not always optimal. In order to choose the level of contemporaneous aggregation at which time series disaggregation should be carried out, the use of formal econometric tests is suggested.     

THE ACCOUNTS AND THE COMPUTER

(1) The primary contribution from the computer's application to the national accounts may well be to erode the line between micro and macro analysis. Key macro totals in the accounts sum individual company reports. The computer permits us to develop distributions of these reports. Such distributions, regularly presented, would permit discovery of the first forerunners of change, would help distinguish, e.g., widespread strength in an export drive or a profits surge, from participation by a few major concerns that dominate the aggregate.
(2) The strikingly different parameters in cross section and time series studies (e.g., price elasticity of housing) will in some measure reflect incomparability between the micro data that enter into each. The computer makes possible the use of the wide array of micro data that really underly the accounts to develop consistent analyses of time series (of both aggregates and distributions) and cross section analyses.
(3) The inconsistencies now imbedded in the accounts but gilded over by the abilities of the estimators are well-known. Discussions of wage price policy rest on data for wages that have no necessary compatibility with data on profits, etc. But since 1,500 corporations account for at least half of U.S. net income, sales, and investment, the computer can test the consistency of reports made by different units in these firms to different agencies—a process totally out of the question before the computer.
(4) The potential that the computer offers for prompt revisions in the accounts; for revisions by systematic rule; for tests of sensitivity of the entire set of accounts to particular tailor-made adjustments, is clear.
(5) Company purchase orders and accounts are increasingly recorded on cards or tapes. From these we may derive input-output detail and process detail that are light years better than those now feasible from intermittent survey aggregates.     

THE HISTORY OF THE UNITED STATES NATIONAL INCOME AND PRODUCT ACCOUNTS: THE DEVELOPMENT OF AN ANALYTICAL TOOL*

The purpose of this article is to record the history of the national income and product accounts of the United States, concentrating on the period 1932–47. During that period the single national income aggregate evolved into a set of accounts and the estimates emerged as an important analytical tool. Interviews with participants in these developments were extensively utilized to trace the events, people, ideas, and other factors which shaped the history of the accounts. The generally recognized need for economic information during the Great Depression stimulated the request that the Department of Commerce undertake what became the first official continuing series on national income in the United States. These estimates were prepared with the cooperation of the National Bureau of Economic Research and were published in 1934. By the late 1930's, estimates were extended to include income by state and a monthly series. World War II was the impetus for the development of product, or expenditure, estimates. By the mid-1940's, the estimates had evolved into a set of income and product accounts–a consolidated production account, sector income and outlay accounts, and a consolidated saving-investment account–designed to provide a bird's-eye-view of the economy. During this period uses of the accounts widened; analysis of wartime production goals and anti-inflation policy are noteworthy examples. The National Income, 1947 Edition was the culmination of a period of intensive conceptual discussion, extension of data sources, and improvement of estimating techniques. Thereafter the mainlines of development are more familiar, encompassing refinement and elaboration of the estimates and proliferation of uses.     

Analyst earnings forecast revisions and the persistence of antidumping relief

Panel corrected standard errors with instrumental variables and effects are invoked to assess the significance of earnings forecast revisions around critical dates in non-steel AD petitions filed in 1985–1987. These petitions were filed between two important US trade law revisions (1984 and 1988), and the period encompasses significant stock market advances and declines. Event studies have been invoked to assess the value of AD petitions. However, they do not estimate the temporal distribution of any abnormal returns. Because analysts make quarterly earnings forecast revisions over several horizons, we can assess the short and long run value of petitions. We find that AD petitions tend to depress earnings forecasts in the year of the petition. However, second year earnings forecasts tend to be revised upwards. There is no effect on five year (long term) earnings growth forecasts. Hence any benefits of protection do not persist. There is evidence that analysts anticipate the filing by revising forecasts in the three months in advance of the filing. We also find that AD petitions do not affect the accuracy of forecasts.     

PROBLEMS OF DETERMINING AND MEASURING THE RELIABILITY OF NATIONAL ACCOUNTS IN DEVELOPING COUNTRIES

Most developing countries have compiled national accounts on a regular basis only for the last few years. It has not yet been possible for them to collect many of the statistics necessary to obtain good coverage of their economic activities by methods which would generally be accepted as reliable. Consequently the checks on reliability imposed by the framework of the national accounts are often absent, and the accounts prepared contain many estimates of doubtful quality. These doubts can usually only be removed as statistics collected by better methods become available. This is proving to be a slow process, partly because of the shortage of trained statistical staff and the competing demands of social and demographic statistics and partly because of the inherent difficulties in collecting good statistics from small businesses and traditional households. The need to define traditional households as producers as well as consumers leads to our demanding extra information from this difficult sector. In addition it is often difficult for the national accounts statistician, and even more so for the user, to find out in the time available exactly how some of the statistics with which he is presented were obtained. When this cannot be done it is impossible to assess their reliability. Thus assessing the overall reliability of national accounts in developing countries for even a limited range of uses is at present largely a matter of personal judgment. The information necessary to make more objective assessments rarely exists and hence the problems which developed countries face in using such information are not yet within the experience of most developing countries.     

AN EXAMINATION OF THE FEASIBILITY OF USING MONETARY DATA FOR NATIONAL INCOME ESTIMATES

In a number of underdeveloped countries today, adequate statistics for estimating national output by traditional national accounting methods are unavailable or unreliable. However, many of these same countries do publish data on monetary variables at an early stage in their development. These data can now be used to estimate national income.
In this study the money supply was defined to include all currency in circulation, private deposits subject to check at all banks and postal systems, all government deposits, and unused overdrafts less float. The national accounts data were taken from United Nations sources and data supplied by various foreign statistical offices. To make the accounts more comparable in terms of coverage and to limit reported income to the monetized sector of the economy, non-monetary imputations were deleted.
The monetary and national accounts data were combined in a multiple, stepwise regression. National income was used as the dependent variable and money supply and other data were used as the independent variables. The final estimating equations explained about 96 per cent of the variation in income between countries. Other tests were conducted using the currency ratio, transactions velocity, population, and per capita consumption. However, these variables did not augment the explanatory power of the regression equations.
When the equations were used to estimate national income for twenty-two under-developed countries, the derived estimates showed a high degree of concordance with reported income where it existed for comparative purposes. The results indicate that monetary data can be used to estimate national income for underdeveloped countries with a relatively high degree of accuracy, between countries, and from year to year within a country.     

THE LEVEL OF WORLD INEQUALITY: HOW MUCH CAN ONE SAY?

This paper constructs estimates of income and consumption inequality for the world (124 countries), using various measures of inequality. It then goes on to examine the possible effects of various sources of error in the estimates, and attempts to set rough limits to the size of such effects. Among the sources of error examined are purchasing power parities used for currency conversion, systematic errors in estimates of per capita incomes, differences in age structure, government tax and expenditure policy, and lifetime income effects. The paper concludes that, although the level of uncertainty in the estimates is too great to permit conclusions about, for instance, trends over time, it is clear that the level of world inequality is extreme, and that it is primarily due to differences in average incomes across countries rather than to intra-country inequality.     

On the estimation of the effect of labour participation on fertility

In this paper we consider the estimation of the causal effect of female labour market status (participation and employment) on fertility. We focus on the sensitivity of the estimated effect to (i) the assumptions about the exogeneity of labour market status; and (ii) the time interval between the measurement of fertility and employment status. Using Spanish quarterly data, we estimate a switching probit model that accounts for the joint determination of both variables. In order to obtain a behavioural effect of the former on the latter, we look at the timing of conception instead of the timing of birth, and present alternative sets of estimates depending on the accuracy with which conception is measured (yearly or quarterly). Our results show a positive although non-significant effect of participation and employment on the probability of having the first child, once the sample of women who conceive in the same quarter (or one quarter later) in which labour market status is measured and the endogeneity between both variables is accounted for. We find that annual data tend to over-estimate the negative effect of employment or participation on the probability of having a child, but the main biases appear when looking at the effect of participation. We are grateful to Adrian Kalwij, Daniel Miles and two anonymous referees for helpful comments on this work. All remaining errors are our own.     

USE OF NATIONAL ACCOUNTS FOR A SHORT-TERM ECONOMETRIC MODEL: 1954–1966

A quarterly macro-econometric model of Japan's postwar economy has been constructed for the period 1954–1965 FY on the basis of standardized quarterly national income accounts. The model is designed for facilitating short-term economic forecasting and formulating adequate fiscal and monetary policy. Longer-term factors such as labor mobility, technical progress, etc., were also considered in the model.
The model consists of fifty-three equations related to most of the macroeconomic variables in both money and real terms, and the equations were estimated in principle by the limited information maximum likelihood method. Principal exogenous variables related to policy instruments are government expenditures including transfers, parameters of tax functions, interest rate, and prices and fares controlled by the government, etc. In formulating the model, non-linear specifications were used whenever found necessary.
Results of our testing on its predictive capability indicated fairly satisfactory performances for our observation period and also for 1966 FY. Multipliers related to fiscal and monetary policy were also obtained, indicating the dynamic characteristics of the Japanese economy, in particular, represented by dynamic business fixed investment, as compared with corresponding multipliers of the U.S. models.
Although the model is exploratory and to serve as a core for a more disaggregated "Master Model," the usefulness of the model for our purposes and the workability of our quarterly national accounts data for model-building have been recognized. The quarterly data, however, still remain to be improved especially in regard to consistency between income and expenditure and integration with flow-of-funds accounts.     

THE SIZE DISTRIBUTION OF PERSONAL WEALTH IN FRANCE (1977): A FIRST ATTEMPT AT THE ESTATE DUTY METHOD

This paper reports upon the first official application of the estate multiplier method of estimating the wealth distribution to French data. It is based upon a sample of estate duty returns filed during the period September-December 1977. The sampling rate was 5 percent for estates under one million francs, and 100 percent for estates over this level, giving a total of 5031 records. The data available did not permit a breakdown by type of asset. It did, however, permit classification of estates by age, sex, and occupation of decedent. Experiments were conducted using five different sets of mortality multipliers. The set of mortality multipliers judged most appropriate leads to an estimate of aggregate net wealth that is 77 percent of that given in the national balance sheet of the national accounts. Comparison of the distributions of wealth derived in these estimates suggest that the figures are consistent with those found in other countries.     

Output gap in Brazil: a real-time data analysis

Recent literature for developed economies has shown that output gap estimates go through important revisions over time, impairing their reliability in real time. We organize a real-time data set for Brazil’s GDP and assess the revisions of the output gap estimated by four different methods. Similar to the findings of studies for developed economies, the output gap revisions in Brazil are substantial in all methods, with frequent changes in the output gap sign. In general, both the GDP data revision and the effect of adding new observations to the sample are relevant sources of output gap revisions. However, differently from those studies, we cannot assert that the latter source is preponderant.     

Why are initial estimates of productivity growth so unreliable?

This paper argues that initial estimates of productivity growth will tend to be much less reliable than those of most other macroeconomic aggregates, such as output or employment growth. Two distinct factors complicate productivity measurement. (1) When production increases, factor inputs typically increase as well. Productivity growth is therefore typically less variable than output growth, meaning that measurement errors will tend to be relatively more important. (2) Revisions to published estimates of production and factor inputs tend to be less highly correlated than the published estimates themselves. This further increases the impact of data revisions on published productivity estimates.To assess the extent of these problems in practice, we detail the importance of historical revisions to the most commonly-used measures of US aggregate productivity growth, expanding on previous empirical work by Aruoba (2008) and Anderson and Kliesen (2006). We find that such revisions have contributed substantially to policymakers’ forecast errors for US productivity growth.     

ESTIMATES OF CAPITAL STOCK BY INDUSTRIES IN THE FEDERAL REPUBLIC OF GERMANY

The main purpose of this paper is to discuss some of the measurement problems in connection with the perpetual inventory method applied for estimates of capital stock. In the Federal Republic of Germany, highly aggregated capital stock data by business sector are compiled by the Federal Statistical Office within its national accounts calculations, while more detailed capital stock estimates by industrial sectors are published by the German Institute of Economic Research (DIW) in Berlin. Because of various gaps in the statistical sources, the accuracy of the capital stock calculations is not yet entirely satisfactory. Aside from the problem of establishing long time series for gross fixed capital formation in constant prices for all sectors, it is difficult to obtain reliable data on the inter-sectoral transactions in secondhand capital goods. In addition, there are problems of determining price indices and service life distributions of the fixed assets in the various parts of the economy. This paper shows a way to arrive at a reasonably close approximation to the latter problem.