vol29-3.indd IASSIST Quarterly Fall 2005 by by Bobray Bordelon* Cross-National & Intergovernmental Data: Paying for One-Stop Shopping Introduction Many intergovernmental and governmental organizations produce their own interfaces for statistical data. With the exception of the Common Database which combines select data from many of the United Nations’ independent agencies, there is typically little cooperation for integrating data between agencies. Some universities have chosen to produce their own interface to merge data into a single source. One example is the Economic and Social Data Service from the United Kingdom. One may also rely on commercial sources. There are various questions to consider. Is a researcher likely to want to combine sources from various organizations? Is it logical to do so? Is the native interface good enough to use on its own and does it is easily allow the combining of data from various sources? Are the data from the original producer easily extractable? Does the original producer allow alteration of the data? Does your organization have the staff and expertise to design and maintain its own interface? Does your organization have the expertise to match changing data elements over time? Is there a commercial service, already in existence, that already does what you want? If so, how do the short-term and long-term costs compare? How would you treat documentation? How does the commercial vendor treat documentation? Commercial Aggregators Since most organizations lack the resources to build its own interface and provide the ongoing maintenance, the choices for a commercial aggregator will be explored. There are many elements to investigate when considering a commercial solution: • Variety of data (due to cost, we want one source to cover as many organizations as possible; ideally it should be international and cover all countries and territories) • Source of data should always be specified • Fixed beginning date or Rolling Period • Start date (should be easy to determine when each series starts) • Frequency of data (daily, weekly, monthly, quarterly, annual) • Update frequency • Downloading capabilities and format (continuous series; text, tab, or comma delimited; EXCEL; SAS, SPSS, STATA) • Documentation (online, direct links, paper) • Interface/ease in finding data • Medium of storage and delivery (INTERNET, WEB, CD, tape, terminal based) • Cost • Customer help (toll free phone number, responsive e-mail, online chat) • Lease or purchase Analyzing the Aggregators Four major aggregators will be examined: Datastream International, EIU World Data, Global Financial Database, and Global Insight. 6 IASSIST Quarterly Fall 2005 Datastream EIU World Data Global Financial Database Global Insight Individual Coun- try Data Yes Yes Yes Limited Sub-national Very little Limited No Yes for USA IMF Direction of Trade Statistics Yes Limited No Yes IMF Balance of Payment Statis- tics No No No Yes IMF Interna- tional Financial Statistics Yes Limited No Yes IMF Govern- ment Finance Statistics No No No No OECD His- torical Statistics/ National Ac- counts Yes Limited No Yes Forecasts Yes Yes No Yes but for an extra fee Exchange rates Yes Yes Yes Yes Interest rates Yes Yes Yes Yes Platform Client Based Web Web Client Based/Web soon Documentation Limited/Extranet Incorporated Incorporated plus Data Encyclopedia Limited extranet for actual series plus more from source for select others (equivalent of a codebook) Another issue to consider is what metadata are provided. Datastream provides the start date, frequency, unit and scale of measurement, adjustment factor, conversion method, and original source. Global Insight and the EIU provide similar information with the EIU also providing the name of the analyst. Global Financial Database provides the most detail and its de� history of an indicator. What is missing? The sources tend to omit actual definitions of the terms being used. The EIU is the only one of the four sources to provide definitions. For the novice trying to choose between several choices, a hyperlink or mouse-over could provide much assistance. It would be useful to provide the prefaces to the works being used or the methodological documentation. For example, including International Financial Statistics Country Notes, Balance of Payments Manual, and Guide to Direction of Trade Statistics would assist the researcher in understanding inconsistencies and subtle distinctions in series. What do we call an item? Even when the same source is used, the vendor may not use the original terminology. The International Monetary Fund tries to standardize terminology between nations by having a consistent vocabulary. Since items are sometimes measured quite differently, the IMF tries to bring together similar concepts by grouping items by line number. For example, line 14a of International Financial Statistics represents the amount of money in circulation outside of banks. The United Kingdom, IASSIST Quarterly Fall 2005 7 Japan, Brazil, and Canada refer to this as “Reserve money of which currency outside DMBs (Deposit Money Banks)”. The United States calls it “Reserve money of which currency outside banks”. The Euro nations, Switzerland, and Sweden refer to it as “currency issued” while Denmark defines it as “Reserve Money of which outside BIs”. Datastream refers to it as “currency in circulation” or “currency outside banks”. Global Insight uses four terms: “currency in circulation outside DMB”, “currency outside banks”, “currency outside banking institutions”, and “currency in circulation”. If this was not confusing enough, one would have to turn to the Country Notes and not the International Financial Statistics Yearbook to determine what constitutes a “DMB” in each nation. A few examples are Qatar considers DMBs to be locally owned commercial banks (including Islamic banks) and branches of foreign banks; Canada includes only chartered banks; and Georgia only commercial banks. In Comoros, it is one specific bank -- Banque pour l’industrie et le commerce-Comores. The confusion continues when a source is listed but the notes provide contrasting information. For example, the EIU provides a definition of M1 for the United States and cites its source as International Financial Statistics. However a note states the data uses national concepts rather than IMF concepts. Throw in similar series having different base dates and one begins to understand why researchers are confused and crave simplicity. What documentation should be provided? At minimum the following should always be listed: • Start date • Frequency • Update Schedule • Formats available for downloading • Unit of measure • Scale • Adjustment factors (if any) • Status (active/discontinued) • Conversion Method (if applicable) • Source • Name of analyst for forecast • Definition • Links to old data if discontinued • Comparability charts How can we help researchers? Talk to vendors about what is needed. Research institutions help pay the bills. Use the collective power of groups such as IASSIST and the American Library Association to persuade vendors to improve their products. Provide guides that point out common mistakes and fallacies. Long ago, Princeton University provided a guide for Datastream when the company itself was providing no documentation. This guide not only provided search examples but also pointed the user to where to look for items and the many misnomers of the program. Provide training to staff and researchers. Have the actual documentation available at a common reference point. While many of the sources are standard monographs that you may not have room for in a reference collection, have the basic guides to documentation and try to keep major statistical works on site. While the user accessing this information off site will not have immediate access, the serious researcher would know where to turn. Perhaps most importantly, we can know the sources’ advantages and limitations and be able to guide the researcher to their appropriate use. * Bobray Bordelon is the Pliny Fisk Librarian of Economics and Finance/Data Services Librarian at Princeton University. bordelon@princeton.edu. The article is based on a presentation delivered at the 31st IASSIST Conference in Edinburgh Scotland on May 25, 2005. mailto:bordelon@princeton.edu