price increase might be put into effect; price analysts were continually suggesting that firms probably would not put all of the authorized price increase into effect but were simply "banking" authorizations in case they needed them in the future.

Second, it was difficult to estimate the universe to which the price increase requests applied. Firms followed different procedures in filing for price increases; some firms filed on specific products and others on broad product lines where cost and thus price increases could be applied to various products. During Phase II, the price increase requests were not initially tabulated by Standard Industrial Classification (SIC) Codes. Even in Phase II or IV, although there was information telling what portion of the sales in a SIC group was affected by a given price increase authorization, there was no price information on the firms in an industry that did not have to prenotify. The relative importance of these non-prenotifying firms varied by industry and with the assumptions one made about price behavior (e.g., price leadership by larger firms).

Third, even if it could be assumed that all authorizations were actually put into place in the market, there was no way of ascertaining the timing of the increases. Price increases might be effected over a long time period, and, worse yet from the standpoint of assessing impact, firms could quickly return with additional price increase requests.

Finally, the actions in one industry could be linked only imperfectly to actions in another. The flow of products through the economy and the effect of different input prices on output prices is only loosely understood. Conceptually, the use of input-output analysis is appropriate in such matters; it is, however, difficult to apply. During Phase II the use of input-output analysis was experimented with but at no time during the entire Program was detailed input-output data available for any time after 1963. Further, even the most detailed I-0 Tables include just 360 sectors—a comparatively small number when one considers making national economic policy.

Thus, from a data analysis standpoint, the regulatory aspect of the Economic Stabilization Program was largely cut-off from the reality observed in the market place. After the fact, as each Consummer Price Index and Wholesale Price Index was released, some linkage of sectors with price authorizations could be made in a rough way. But, with the possible exception of an automobile price increase or a steel price increase, it was difficult to approximate, with any confidence, the effect on observed prices of a given regulatory decision or a given policy change.

The need for analyzing the economy and the relationship of price regulations to the economy was most evident during the decontrol considerations of Phase IV and during the planning of a followon agency to the Cost of Living Council. During decontrol, several questions were continually raised: What is the demand situation vis-a-vis capacity in the industry? How is the Economic Stabilization Program affecting investment within the industry? What will happen to prices in the absence of direct controls? Have controls led to a large increase in exports of critical materials?

One of the first things to surprise people who are not familiar with government statistics is that there is little information about prices, let alone the determinants of prices. There are, of course, some commonly available indices of prices-the Consumer Price Index, the Wholesale Price Index, and the GNP deflator. These indices are also disaggregated into a series of sub-components. However, even in terms of coverage, these indices are inadequate for the analysis of price determination. The WPI is available for some SIC breakdowns, but this breakdown is not comprehensive. Further, because of the small number of commodities actually priced (sampled) and because of the use, in some cases, of secondary sources or of list prices instead of transactions prices, there is serious question about the usefulness of these indices during cyclical periods. In terms of individual commodities, it is virtually impossible to track prices over time.

The problems with price data is repeated in other critical areas. The Federal Reserve Board and the Wharton School/McGraw Hill construct aggregate capacity indices for major sectors of the economy. However, these capacity indices do not give an accurate portrayal of the level of operation of any given industry because of the absence of any benchmark data (i.e., what is "full capacity"?) and because they are so aggregated that they may often miss the detail that is important for spotting trouble while it is brewing. In the international trade, the situation is slightly better. While there is little available price data, there are trade statistics on many commodities. But international trade data are not reported until three months after the trade occurs, making it difficult to react to changing situations.

This discussion is directed at the difficulty in simply tracking data about key elements of the economy. Obviously, data must be available before it is possible to understand the underlying causal relationships with regard to price movements, investment, etc., within the economy. However, for developing long-run stabilization (and

1 One should note that new and hopefully better indices of capacity are being planned by the Federal Government.

data) policies, it is necessary to analyze these underlying relationships.1

Phase I Data Systems

August 15, 1971 to November 14, 1971

The Phase I Freeze was basically dependent on voluntary compliance. Consequently, during that period, no regular reports were made by commercial entities to the Cost of Living or its field operational arms (the Office of Emergency Preparedness and the Internal Revenue Service) and no data base of economic interest was developed.

There were, however, two computerized data bases developed as part of Phase I. The first of these contained rulings on exemptions, exceptions, and answers to general questions concerning the interpretation of Freeze regulations. This data was stored in a real-time information retrieval system and was available to the Office of Emergency Preparedness regional offices by means of key-word search on remote terminals. From OEP regional offices, this information was distributed to IRS district offices.

The second data base contained information on alleged violations and inquiries. All complaints and inquiries were encoded on a form, readily entered into the computer, designed at the specific direction of Arnold R. Weber, then Director of the Cost of Living Council, by Conrad Jacobs and Benjamin Kaplan, who were detailed to the Cost of Living Council from the Bureau of the Census. The form was structured to give information on the complainant, the alleged violator, the type of violation (whether one of price, wage, or rent) and the specific type or area of violation within the preceding three major categories. Similar information was gathered on inquiries and inquirers.

The data from these forms were tabulated by each IRS office on a daily basis. The information was then reported to the appropriate IRS district offices. From there it was transmitted to the OEP regional office where it was entered by remote terminal to the central system in Washington. This made daily reports possible, which contained broad tabulations on the nature of compliance or non-compliance with the Freeze, the degree of public acceptance and understanding, and geographic or major Freeze item "hot spots" (e.g., a great number of rent violation complaints in the New York City area could call forth a special compliance strike). On the other hand, this

1 See Appendix I for a discussion of the data requirements of a monitoring agency as they relate to the submission of company data to that agency.

daily tabulation at the IRS offices took many man-hours from violation or inquiry handling to fulfill these reporting requirements.

Both of these bases were entered on the EMISARI system, which was designed by Dr. Murray Turoff of OEP as part of the regular OEP need for quick communication and information acquisition during disaster situations. The EMISARI system ran on a Univac 1108 computer with a specially concatenated core memory. This computer was located in the Washington area and was already fully available for OEP use prior to the Freeze as was the basic software of the EMISARI system. Since this software system only had to be modified for Freeze use and the violation and inquiry form was designed overnight, the data bases and reporting systems were up within a week of the initiation of the Freeze.

There was also a secondary computer system set up to tabulate, store and retrieve the material from the Alleged Violation Report forms. The forms were sent directly from the IRS offices, by mail, to Jeffersonville, Indiana where, at a local office of the Bureau of the Census, the data were key punched, transmitted to Washington electronically, and entered, in batch, on a Census IBM 360 system. The design of the computer programs that created the administrative reports from this data required merely ten days from start to final debugging.

This Census system and the EMISARI real-time violations and inquiries data base were, in part, obvious redundancies. The EMISARI system had the advantage of daily updates, but was extremely burdensome to the IRS and OEP regional staffs and provided less detail than the Census system. The Census system was also more accurate in that a key punch error on the entering of one form in Jeffersonville would not have the same deleterious effect on the data base as a key board error in entering the aggregate number from one IRS district at an OEP regional office terminal. As it were, both systems ran through Phase I and into Phase II, where the EMISARI and Census systems were retired and both replaced by IRS facilities.

1

A summary record of this Phase I data with interpretations is provided in OEP's history of the Freeze and Arnold Weber's account of his Freeze experience 2 so it will not be reproduced in the data appendix along with the Phases II, III, and IV data. Although Phase I data are not particularly interesting in an analytic sense, Phase I data systems anticipated some of the key problems of data

'Harry B. Yoshpe, et. al., Stemming Inflation: The Office of Emergency Preparedness and the 90 Day Freeze (G.P.O., 1972).

2 Arnold R. Weber, In Pursuit of Price Stability: The Wage-Price Freeze of 1971 (Brookings, 1973).

management in later and more complex phases of the Economic Stabilization Program: (1) what data to collect; (2) centralized or decentralized data entry and edit; (3) real-time or batch mode operation; (4) using a predesigned computer software system or creating a new system; (5) quality control of the data base; (6) redundancy of systems and data bases; and (7) the uses of the data system output.

Phase II Data Systems

November 14, 1971 to January 11, 1973

During Phase II the Price Commission (PC) operated four data systems concurrently: an "interim" computer system by Interactive Data Corporation (IDC) designed for immediate needs, a system designed by Arthur Andersen and Company to meet longrange need, a decision data system jointly implemented by economists at the Price Commission and the Cost of Living Council to serve analytical requirements, and a manual system developed within the Price Commission's Office of Program Operations (OPO) to produce management reports. The four systems processed essentially the same source documents and, to a large extent, provided the same information for the same group within the Price Commission. In addition, the Internal Revenue Service (IRS) adopted the Phase I violations reporting system to fit the needs of Phase II.

The Price Commission expected to receive several hundred forms daily from different sources, all of which needed computer tracking. During Phase II, companies with annual sales of more than $100 million had to file the PC-1 form to prenotify price increases 30 days in advance of intended implementation and to cost justify these increases. In addition, there were PC-50 and PC-51 forms for quarterly reporting of sales, costs, and profits by firms with sales in excess of $50 million; PC-5 forms for reclassification of requests by multi-industry firms; PC-10 forms for reporting markups by retailers and wholesalers; and PC-35 forms for public utility price increases. The Commission also received many of the forms which they had directed the IRS to institute. These included S-53, a supplemental worksheet for non-institutional health service providers, and the S-16, an application for exemption or exception.

In addition to tracking a multitude of forms, the Price Commission believed that a data base would be needed to support policy decisions. The following functions of a computerized data base and management information system were identified:

1. To act as a case tracking system which would monitor the flow of documents, identify bottlenecks and workloads, and