PRINTER-FRIENDLY

About Recovery Act Data

• Recovery Act Recipient Reports Data
• Recovery Data Dictionary

About FedSpending.org Data

• FedSpending.org Data (Contracts and Assistance tabs)
• Contracts Data Dictionary
• Federal Assistance Data Dictionary

Recovery Act Recipient Reports Data (Recovery tab)

The data presented on the Recovery tab of FedSpending.org are downloaded from Recovery.gov, the website created by the American Recovery and Reinvestment Act (Recovery Act), and have not been altered. Except for the additional of some data fields, they are identical to what the public can download as recipient reported data from Recovery.gov.

The Recovery tab of FedSpending.org allows searching of over 100,000 reports from prime recipients of more than $250 billion in contracts, grants, and loans awarded under the American Reinvestment and Recovery Act (Recovery Act) between Feb. 17, 2009 and June 30, 2012, as well as additional information on sub-recipients and vendors. Recipients of Recovery Act funds are required to report quarterly about use of the money they receive. The first reports were made public on Oct. 30, 2009. Fedspending.org uses a dataset that was last updated on Sep. 19, 2012.

Users can search through these records - individually or in aggregate - for a particular prime recipient, sub-recipient, ZIP code, congressional district, federal agency, award amount, and more.

In some cases, FedSpending.org presents or processes data on the Recovery tab in a different way than it is shown in the original government database. This is generally done for the convenience of users so data is more understandable and accessible. The original data fields and information from the government are also available unmodified in the Recovery tab of FedSpending.org and may be seen if the user chooses the "Recipient Reports" level of detail option available on all data searches in the Recovery tab. For more information on data fields in the Recovery tab, see the Recovery Data Dictionary.

"Net Amount Retained"

While the Center for Effective Government has not changed the data from Recovery.gov, we have added a new field, called "Net Amount Retained." The Net Amount Retained is the amount of money a prime recipient or sub-recipient does not pass on to another entity (such as a sub-recipient or vendor). This field is used to show the extent to which Recovery Act funds are passed from the prime recipient to a sub-recipient or a vendor without double-counting funds in the totals for searches. FedSpending.org's Recovery tab includes the "retained" calculation because it can be useful for understanding the actual amount of Recovery Act funding that stays with a certain entity or at a certain location.

For example, a search (see below or here) for how much money went to recipients in Arizona shows that prime recipients in the state were awarded $4.3 billion; they provided $1.1 billion in sub-awards and $235 million to vendors for a net amount retained of $3.0 billion. Note that some of the sub-awards or payments to vendors might be outside of Arizona, so they don't match the independent totals within the search for sub-recipients and vendors located in Arizona. (The website provides different levels of detail to learn more about where the money is going.)

Sub-recipients in the state received $1.1 billion; they provided $2 million to vendors (the government is not collecting information on the amount sub-recipients provide in sub-awards) for a net amount retained of $1.1 billion. Vendors located in the state received $12.7 million from both prime and sub-recipients. Adding each of the net amounts retained plus the payments to vendors gives the total net amount of Recovery Act funds retained by Arizona recipients, which is $4 billion for this search.

Cumulative vs Quarterly Datasets

The Recovery.gov Web site now supplies, and Fedspending.org uses, a cumulative dataset as well as a quarter-by-quarter dataset. The basic purpose of the cumulative dataset is as follows: recipients of Recovery Act funds must report in each quarter. Each quarter's report contains the full award amount they have been awarded to date, the full amount they have received, and so on -- not just the amount awarded or received in that quarter. Therefore, adding recipient reports from more than one quarter together in a straightforward way would result in double or multiple counting.

However, you can't just use the latest quarter information by itself if you want a full picture of Recovery Act funds, because it won't include recipients whose project completed in earlier quarter(s) and therefore didn't report in the latest one. The cumulative dataset includes records from all quarters that can be added together in a straightforward way to get a total of Recovery Act funds.

To quote a prior version of the Recovery.gov site from when the last quarter ended March 2010, the cumulative dataset "include[s] data on: awards made from January 1, 2010 ? March 31, 2010 that recipients reported on in April 2010; projects completed prior to March 31, 2010; and projects not completed prior to March 31, 2010 that the recipients should have reported on again in April 2010, but did not."

We believe that most users of Recovery Act data will want to use the cumulative dataset. The cumulative dataset is what has been used to prepare all of the tables that are accessed through the Fedspending.org sidebar, and is the default for all searches. However, users who would prefer the quarterly dataset (because they wish to which reports actually came in in which quarter) can select that instead through the Advanced search. Because it would be highly confusing to add together records in the quarterly dataset across quarters, you can only see one quarter of data at a time if you do this.

Records in the cumulative dataset have apparently been assigned to the latest quarter by Recovery.gov if the award that they are for extends into that quarter. The only records for earlier quarters are apparently only those that ended in that quarter.

Choice of which quarter you would like to retrieve through the Advanced Search works along with cumulative vs quarter dataset as follows:

1. If you do a cumulative search with no quarter choice, you get a cumulative total of all the Recovery awards so far (that meet the other criteria of your search).
2. If you do a cumulative search and choose a quarter, you get either:
   1. All of the projects that finished in that quarter, if you chose a quarter that is not the latest one, or;
   2. All of the projects that either started in the quarter or continued into it, if you chose the latest quarter.
3. If you do a quarterly search, you must choose a quarter or it defaults to the latest one. You get whichever projects were reported on in the quarter that you chose.

A Note on Data Quality

As noted above, the data on the Recovery tab of FedSpending.org were reported by recipients of Recovery Act funds and have not been altered by the Center for Effective Government. The site to which recipients report, FederalReporting.gov, performs minimal checks on the information to ensure that recipient reports are accurate. Recipients of Recovery Act funds are allowed to enter questionable (for example, zero jobs reported created or saved for projects expending more than $1 million) and even nonsensical (for example, entering spending reports for congressional districts that do not exist) data. Currently, Recovery.gov and FederalReporting.gov have no mechanisms to screen for these data.

An example of poor data quality can be seen on the summary view for the entire FedSpending.org Recovery tab dataset (see here). The amount awarded to sub-recipients by prime recipients should equal the amount received by sub-recipients from prime recipients. However, the method by which we calculate both amounts indicates that these two figures are not equal. This inequality is the result of sub-recipient reports containing award amounts different from what is indicated in prime recipient reports and errors in other data that link prime recipient reports to sub-recipient reports.

It is important to note that one of the reasons transparency is important is because it serves to improve data quality. That millions of people have seen these reports and call into question some of the data is the first step toward implementing policies and technology that will increase the quality of data reported by recipients of federal funds.

Background on Recovery Act Recipient Reporting

On Feb. 17, 2009, President Obama signed into law the Recovery Act, providing $787 billion to boost the economy and address the nation's health, education, energy, and infrastructure needs. Most of the money will be spent by 2012, with the largest amounts being spent in 2010.

President Obama promised unprecedented levels of transparency and accountability for this law. Before signing it, the president said every American would be able to easily track the spending online and announced the launching of Recovery.gov. The Recovery.gov website provides public access to roughly one-third of the $787 billion in Recovery Act funding; it only includes information on federal grants, loans, and contracts, and nothing on entitlements or tax breaks.

This is the first time that recipients of federal funds must report in a timely and transparent manner. While this new approach has potential to set a precedent for reporting and disclosure under annual federal spending, the first round has also been fraught with data quality issues. Since the data on this website comes from Recovery.gov, it includes whatever data errors that website contains.

FedSpending.org Data (Contracts and Assistance tabs)

Duplicate records and IDs in data from usaspending.gov

There were 13,169 pairs of records in FAADS, the assistance database, with the same unique_transaction_ids -- these are IDs that are supposed to uniquely identify each record, but did not in these cases. Of these records, 116 were from various agencies, but the large majority of them were loan records from the Export-Import Bank. Checking some records on usaspending.gov showed that only one of each pair of records was displayed. In addition, the pairs of records were mostly identical to each other within each pair, except for a few data fields. In particular, the recipient category data field contained the code translation as well as the code for half the records, and just the code for the other half.

Since half of these apparently doubled records do not seem to appear on usaspending.gov, fedspending.org has deleted the doubles from its database, deleting the record of each pair whose recipient type category was not translated. This results in the unique_transaction_ids in FAADS being actually unique once again. We tried to contact usaspending.gov about this but were unable to obtain any more information on this problem.

Unusual Datapoint in FY2009 Data Load (added November 2, 2009)

There was some unusual data contained within the data load launched on October 22, 2009 related to the "Construction of structures and facilities" product and service code. Specifically, FedSpending.org currently shows the federal government to have deallocated just under $8 billion in FY 2009 for that product and service code (see screenshot below). While deallocations are present throughout the FedSpending.org dataset, it is very unusual to have such a large deallocation.

After investigating the data further, it appears a single contract transaction to EMERSON CONSTRUCTION COMPANY, INC. was incorrectly entered into the Federal Procurement Data System (FPDS) on September 29, 2009 (the second to last day of FY 2008). This transaction was for over $13 billion, or 62 times more money than Emerson had received in all of its federal contracts totaled up since FY 2000. It was also roughtly same amount of money the federal government spent on construction of structures and facilities in FY 2003, which it spread across 5,224 different contractors that year.

We suspect the error was discovered and corrected at a later date, after FY 2008 had ended, because Emerson is listed as having an almost equally large deallocation in FY 2009 on January 7, 2009. Our experience has shown it is typical when correcting errors in FPDS to simply add another transaction that reconciles the total amount given to a contractor rather than editing the actual transaction that contained the error. Unforunately, the end result for FedSpending.org (and USAspending.gov for that matter) is that the totals for this product and service code in both FY 2008 and FY 2009 are incorrect, as are totals for other views on FedSpending.org that included those two transactions to Emerson Construction Company.

Data Modifications

In some cases, FedSpending.org presents or processes data in a different way than it is shown in the original government databases. This is generally done for the convenience of users so data is more understandable and accessible. The original data fields and information from the government are also available unmodified in FedSpending.org and may be seen if the user chooses the "Complete (all information)" level of detail option available on all data searches (For more information on the complete levels of detail fields in FedSpending.org, see the Contracts Data Dictionary and the Federal Assistance Data Dictionary.

The cases where FedSpending.org has modified data from the Federal Assistance Awards Data System (FAADS) and the Federal Procurement Data System (FPDS) are outlined below in three sections:
Changes to Assistance and Contracts Data
Changes to Assistance Data
Changes to Contracts Data

Contracts and Assistance Dollar Quantities

Fedspending presents dollar amounts of contracts and assistance as totals of obligated amounts (for contracts) and Federal assistance amounts (for assistance), except in one case -- for assistance amounts for loans, the Original Loan Subsidy Cost is used instead of the Federal Assistance Amount. This is done to match Usaspending.gov and follow OMB guidance.

In data releases before this latest one, Fedspending presented Award records but not IDV (Indefinite Delivery Vehicle) records, records which can allocate money to be spend in one or many later awards. Fedspending now includes IDV records. These are added into totals via their obligated amounts, just like awards are. There are additional data fields for IDV records which give the total amount that may be spent over the life of the vehicle; these have not been totalled since this money may not actually be spent.

Changes to Both Assistance and Contracts Data

Congressional District
We have regularized Congressional Districts to some extent by treating all incorrect codes/districts within a state as being district "90" (unknown), "98" (nonvoting), or "00" (statewide), depending on the state. Records with invalid state codes are listed as district "ZZ".

Major agencies
FedSpending.org has modified FAADS, FPDS, and Recovery data to streamline the display of federal government agencies. Rather than produce tables that include each sub-agency within the data, it is advantageous to be able to produce tables and output by "major" agency. These generally correspond to departments of the federal government, but some were also chosen because they have a large number of awards. The "major" agencies are shown below:

*The Dept. of Defense, as listed here, includes the Dept. of the Army, Dept. of the Air Force, Dept. of the Navy, the civil Corps of Engineers, and the United States Soldiers' and Airmen's Home, even though these have different initial two-digit agency codes in the federal agency coding scheme.

**GSA is listed as a major agency for contracts, but not for assistance. NSF, NRC, U.S. AID, and the Office of Personnel Management are listed as major agencies for assistance, but not for contracts. Otherwise the FAADS and FPDS major agency lists are the same.

***The agencies combined into "All other agencies" above are different between FAADS, FPDS, and Recovery data. Below are details for the "All other agencies" category for each database:

FAADS "All other agencies"::

FPDS "All Other Contracting Agencies":

FPDS "All Other Funding Agencies":

Recovery "All Other Awarding Agencies":

Recovery "All Other Funding Agencies":

Changes to Only Grants Data (FAADS)

Recipient Definition and "Best Recipient Name"
FAADS now includes a recipient Dun & Bradstreet (D&B) number that can be used to define a single recipient of assistance across multiple records (i.e., one entry that would list multiple awards for the same recipient). It is convenient to group records together in this way, since it is useful to look up a single recipient and find its total assistance. However, not all recipient records have D&B numbers.

Therefore, FedSpending.org has created its own recipient ID based on the following steps: first, records with the same D&B number are grouped together. Then, those without a D&B number have the name of the recipient in each record regularized by discarding some punctuation, removing words like "Co." or "Inc.", and standardizing certain abbreviations. Then, all records with no D&B number and with the same regularized recipient name and recipient state are assumed to belong to the same recipient. Recipient groupings are made in this manner across multiple years.

This method of grouping records with no D&B number by recipient name and state has some disadvantages. It may, for instance, treat a multi-state recipient as being more than one entity, thereby listing the same group twice in different states, or it may group unrelated entities together that happen to have the same name within the same state. But the advantages of a common recipient name to enable aggregate totals across records outweigh the problems.

Grouping records together by D&B Number means that a single recipient may now exist (via scattered facilities or offices, presumably) in more than one state. Therefore, when recipients are listed, the recipients that are defined by D&B number have their D&B number listed, while the ones whose records had a blank D&B number have their state listed.

After records are grouped together by recipient, a single overall recipient name is chosen for each group of records, based on the most commonly occurring name within the un-standardized recipient names for the group. In other words, if a group of three records in the same year all had the same D&B number, but two records had the recipient name Alpha Corp while the third record had the recipient name XYZ Corp, Alpha Corp would be used as the overall recipient name for this recipient. Fedspending.org refers to this as the "Best Recipient Name", and the original recipient name provided as part of each record as the "Recipient Name". Most name searches on the site use the Best Recipient Name, because this guarantees that you'll retrieve all of the records for a recipient when you search by name. However, you can search on the original Recipient Name if you prefer by using the Advanced Search.

Fiscal Year
Sometimes FAADS marks information as being either corrections to data for a previous fiscal year or as late filings for a previous fiscal year. To give the proper scope of federal spending in any given fiscal year, we have displayed these data as being part of the fiscal year that they modify, not as part of the fiscal year in which they were reported.

Recipient Type
FAADS' categorization of what type of entity receives assistance (e.g., individual, state government, private higher education) and how the assistance is delivered (e.g., direct payment, loan, block grant) includes too many values to be conveniently displayed on FedSpending.org. Therefore, we have grouped the original recipient types and assistance type into streamlined summary categories. Users can access the detailed categories FAADS uses by selecting the "Complete (all information)" level of detail option available for all data output. The FedSpending.org groupings for recipient type and assistance type are detailed in the two charts below:

Recipient Type Translations

Assistance Type Translations

Catalog of Federal Domestic Assistance (CFDA) Program Titles
The Catalog of Federal Domestic Assistance (CFDA) Program is a database that classifies all federal programs that provide funding to local government agencies, private institutions, and individuals. Each program is assigned a number and name. Most of the time, CFDA program numbers match up with associated program titles easily. However, there are instances of CFDA program numbers appearing in FAADS that are not listed in the current CFDA program title list. When this happens, FedSpending.org chooses a single best program title for each non-associated number. In some cases this suppresses variations in program titles for a single CFDA program number.

Changes to Only Contracts Data (FPDS)

Parent Company Definition and Names
The most basic task that many users want to accomplish is to look up a particular contractor and find out which contracts they have received. This makes the most sense if done by parent company -- the entity that owns multiple divisions, "doing business as" names, or subsidiaries that may show up as individual contractor names.

FPDS now includes a parent company name and D&B (Dun and Bradstreet) number. It also includes a set of vendor names (ordinary name, alternate name, doing-business-as name, legal name), and a vendor D&B number. Unfortunately, some of the parent company D&B numbers are not filled in, and some of the parent company information varies within D&B number -- for instance, a group of records may have the same parent D&B number but different parent company names.

It is clearly adventageous for users if records that are owned by the same business entity can be grouped together under a single name. Therefore, Fedspending created additional name fields as follows: first, a "best vendor name" was created for each record with the vendor name, alternate name, legal name, doing business name, or parent name in it -- whichever was first filled in. In other words, if all of the vendor name fields for the record were blank, the parent name would be used as the best vendor name.

Then records were grouped together by parent. This was done by 9-digit parent D&B number, if it existed, by vendor D&B number if parent D&B was blank, and by regularized parent name if both D&B numbers were blank. In addition, records with blank parent D&B numbers were assigned the parent D&B number of other records that had one if those other records had the same vendor D&B number, and at least half of the records in the database with that vendor D&B number had that parent D&B number. Parent assignments made in this way were themselves modified by hand for some large companies to assign vendors that were in the same parent company together.

Each grouping of records made in this way was assigned a parent ID, and then a single parent name for the group was created by having a computer program select the most common parent name within records in the group. Fedspending calls this most common name the "best parent name" and displays it in most places where a company name is displayed. Groupings were made in this way across multiple years.

Competition Category
There are five data fields in FPDS that are useful for assessing the level of competition of a contract: the Extent Competed, Reason Not Competed, Number of Offers Received, Statutory Exception to Fair Opportunity, and Type of Set Aside. FedSpending.org has combined information from all five of these fields into one overall summation of the amount of competition for each transaction, called the competition category. More specific "transactions" are summed by these new competition categories in various tables.

Transactions were classified as follows. "Full and open" competition means that all responsible sources (or potential contractors) are permitted to compete. Those with Extent Competed indicating full and open competition or competitive delivery order were provisionally put in the full and open competition category. "Full and open with at least one exclusion" includes competitions in which a certain sources are not permitted to compete. "Exclusive competitions" can happen only if certain conditions are met. Transactions were removed from this category that had a reason for non-competition, a statutory exception to fair opportunity, or a set-aside and added to the "competition after exclusion of sources" category. Those remaining in the full and open competition category that had zero or one offers were moved to a category of full and open competition with only one bid. The other noncompetition categories are modeled from FPDS Extent Competed values. "Follow-on contract" is a contract that continues the funding stream and obligations of a contract that has already been issued.

The table below shows how FedSpending.org streamlines FPDS competition categories:

Competition Graphs
Contractor profiles and summary level search results receive pie charts which graphs four categories for levels of competition. The graph uses Full and Open Competition, and Full and Open Competition But Only One Bid unaltered from the rest of the site. However, most of the remaining categories (Competition after Exclusion of Sources, Follow-on Contract, Not Available for Competition, and Not Competed) are compressed into one group on the graph (although the accompanying table still lists the specific amounts for the individual categories) which represents less than full competition. The final category used for the graph is unknown. It should also be noted that for graphing purposes, negative numbers, which represent deobligated funds, are treated as zero in the graph (although they are listed in the accompanying table).

Contracts Data Dictionary

This is a list of all of the data fields in FedSpending.org's copy of the FPDS (federal contracts) database. It can be used to see what kind of data each field displayed in the extensive or complete levels of detail contains. The "Field Type" column provides the type and size of the field, in case you wish to load FedSpending.org data into your own database. The "% blank" column shows what percentage of the data records have the data field blank -- this can be useful when you want to estimate whether a particular data element is well maintained or not.

Section: Amounts

This section shows dollar amounts of awards and IDVs.

Section: Purchaser Information

This section shows information about the purchaser for the award or IDV, i.e. the governmental agency or agencies that made the purchase, and what programs it was purchased under.

Section: Contract Information

This section shows basic contract information for an award or IDV.

Section: Vendor Information

This section shows basic information about the contractor/vendor for an award or IDV, such as its name, address, parent, and contact information.

Section: Principal Place of Performance

This section shows the principal place of performance of the award or IDV -- the place where most of the work was done.

Section: Product or Service Information

This section shows information about the product or service purchased.

Section: Record Information

This section shows information about a FPDS record, such as its identifying IDs, its fiscal year, its version, etc.

Section: Competition Information

This section shows information on how an award or IDV was competed.

Section: Vendor Characteristics

This section shows characteristics of the contractor or vendor for an award or IDV, such as its size, its type, and a set of flags showing whether it is in various special categories.

Section: Legislative Mandates

This section shows whether various legislative mandates apply to the award or IDV.

Federal Assistance Data Dictionary

This is a list of all of the data fields in FedSpending.org's copy of the FAADS (federal assistance) database. It can be used to see what kind of data each field displayed in the extensive or complete levels of detail contains. The "Field Type" column provides the type and size of the field, in case you wish to load FedSpending.org data into your own database. The "% blank" column shows what percentage of the data records have the data field blank -- this can be useful when you want to estimate whether a particular data element is well maintained or not.

Section: Recipient Information

This section shows information about the recipient of the award, such as the name, location, and type of the recipient.

This section is displayed in All detail reports.

Section: Project and Award Info

This section shows information about which agency granted the award under what governmental program, and provides some ID numbers for the award.

Section: Action

This section shows a detailed action taken under the award -- an amount of money granted or subtracted, the dates involved, the fiscal year, and the type of action.

This section is displayed in All detail reports.

Section: Principal Place

This section shows the principal place of performance of the activity supported by the award.

This section is displayed in All detail reports.

Recovery Data Dictionary

This is a list of all of the data fields in FedSpending.org's copy of the Recovery database. It can be used to see what kind of data each field displayed in the recipient reports level of detail contains. The "Field Type" column provides the type and size of the field, in case you wish to load FedSpending.org data into your own database. The "% blank" column shows what percentage of the data records have the data field blank -- this can be useful when you want to estimate whether a particular data element is well maintained or not.

Section: Recipient / Vendor Information

This section gives basic information about a prime recipient, sub-recipient, or vendor.

Section: Basic Award Information

This section gives basic information about the award, such the date of award and various ID numbers.

Section: Award and Job Amounts

This section shows the amount awarded, the number of jobs created, and other related quantities.

Section: Funding Source

This section gives information about the funding source of the award -- the governmental agency or agencies involved.

Section: Project and Award Description

This section provides a description of the project and the various activities undertaken under the award.

Section: Place of Performance

This section gives the location of the primary place of performance of work done that was funded by the award.

Section: Infrastructure Contact

This section provides contact information for a contact person for infrastructure related to the award.

Section: Recipient Officer Compensation

This section provides information about highly compensated officers of a recipient of an award.

Section: Recordkeeping Information

This section contains basic recordkeeping information about the award record, such as timestamps, data quality flags, and a flag and justification for late submission.

ABOUT THE CENTER FOR EFFECTIVE GOVERNMENT | ABOUT THIS SITE | SITE MAP | CONTACT US