How common are direct-method cash flow statements?

A tutorial on working with XBRL data compiled using dera.pq

Author

Ian D. Gow

Published

11 June 2026

Modified

23 October 2026

Abstract

This note illustrates how data compiled using my dera.pq R package can be used to analyse XBRL data from SEC filings. Using these data, I show that direct-method cash flow statements have become even rarer for US firms than they were in 2011, when the last Accounting Trends and Techniques survey was published.

PDF version (11 June 2026)

1 Introduction

Many filings with the Securities and Exchange Commission (SEC) are tagged using XBRL (eXtensible Business Reporting Language), a global framework for encoding and sharing business information. The SEC’s Division of Economic and Risk Analysis (DERA) provides bulk files compiled from these filings and my recently published R package, dera.pq, is designed to facilitate fast analysis of these data sets. This note provides a case study of how data compiled with the dera.pq can be used and provides a kind of tutorial on parts of the XBRL data set provided by DERA.

The specific question I explore here is: How many US firms provide direct-method cash flow statements? Accounting Trends and Techniques, an AICPA publication, was long the source for an answer to this question and it typically reported that about 1% of firms used the direct method. However, the effective demise of Accounting Trends and Techniques in 2011 has deprived us of an easy answer to this question.

Can we get an answer from XBRL filings? Well, yes. While it is surprisingly difficult to corral the messy, error-ridden XBRL data provided by individual filers into a clean answer, it seems that the headline 1% number is out of date. During the period I study, the popularity(?) of direct-method cash flow statements has been in steady decline and today perhaps 0.2% of firms use direct-method cash flow statements.

Despite their rarity, the belief that direct-method cash flow statements would be useful to investors seems to persist. As recently as December 2023, the Chief Accountant of the SEC issued a call for more firms to use the direct method. Interestingly, even the SEC Chief Accountant appears to rely on statistics from no later than 2002 to describe the state of play regarding direct-method reporting by firms. So perhaps more recent data would be helpful.

1.1 Some observations on the SEC’s XBRL data

While my initial goal with this note was the illustration of how data collated with my dera.pq package can be used, a side effect of of preparing this note is that I am able to make some observations on the utility of the XBRL data based on my experience. While these observations are based on my limited use of the data in preparing this note, I conjecture that they may hold up in a more comprehensive analysis.

First, I view the XBRL data as having greatest utility when analysing longer time periods or comparing larger samples of firms. In April 2025, the SEC published some Python code illustrating the use of DERA data. However, these examples typically quickly move to looking at particular fields for a particular firm for a small number of periods (say, two quarters). If that were the goal, there is arguably little value in using XBRL at all; one can quickly go to the relevant filing and view the few required numbers in context and quickly type them into Excel.1 But the approach suggested in the SEC code would quickly become unworkable for the kinds of analyses for which having XBRL would be more useful.2 In this regard, the dera.pq package provides a better approach.

Second, the minimum quality of the XBRL data is disappointingly low. For bulk analyses, it is probably the lower bound of data quality that is most relevant. Having 90% of the data above a certain quality does not help with comprehensive analyses if the other 10% cannot be ignored. If the bad 10% needs to be included, then a lot of work is needed to patch it up, likely much more than dealing with the better 90%. Even then, one’s confidence in the resulting data will be lower than if greater assurance had been built into the process of producing it.

While I use the data in a very limited way in this note, data issues plague my analysis.3 Many firms do not provide the data explaining how various data points are aggregated and organized in a way that can be understood by a computer. On top of this, there are too many tags provided by the US GAAP taxonomy and many firms use custom tags, seemingly without good reason. I have seen many incorrectly tagged items in my limited use.

Finally, while one might not choose the general approach of tagging used in XBRL if starting with a clean sheet of paper, I think that the fundamental technology is perfectly workable. Harris and Morsfield (2012) suggest that XBRL was out of date even in 2012. However, I have found working with the XBRL data in the form created using the dera.pq package based on the files supplied by DERA to be fast and facile. The much bigger problem is that the quality of the data is too low.

Given how long the XBRL project has been around and how poor the data quality appears to be, I must say I would assess a low likelihood that the XBRL data in SEC filings will be improved any time soon, if ever. The solution seems straightforward in principle: there should be some assurance around the XBRL data in terms of how the items are tagged, how they relate to each other, and how taxonomies are used. This is a solution that would have been viable fifteen years ago and clearly the decision was made not to take XBRL data seriously. As such, it’s unclear that anything fundamental has changed to make it more attractive to invest time, money, and political capital in improving the quality of data that surely few really use.

Even in our current AI age, this seems to be a shame, because I conjecture that what AI can do with high-quality curated data sets is quite different from what it can do if dealing with messy, dirty data of the kind currently seen in the XBRL data set provided by the SEC.4

2 Background

2.1 Cash flow reporting under US GAAP

US GAAP has required most public firms to provide statements of cash flows since FAS 95, which was issued by the Financial Accounting Standards Board (FASB) in 1987. When the FASB created the Accounting Standards Codification (ASC) in 2009, the substance of FAS 95 became ASC Topic 230, Statement of Cash Flows.

ASC 230-10-45-24 requires a statement of cash flows that reports net cash provided or used by operating, investing, and financing activities. In effect, the statement of cash flows reconciles beginning and ending balances of cash and cash equivalents.5

ASC 230-10-45-25 says (emphasis added) “entities are encouraged to report … net cash flow from operating activities [using] the direct method.” 230-10-45-30 requires that a firm that “uses the direct method” provide a “reconciliation of net income to net cash flow from operating activities shall be provided in a separate schedule.”

2.3 XBRL data

With the demise of Accounting Trends and Techniques, it is not clear that anyone is compiling statistics on the prevalence of the indirect method for reporting cash flows from operating activities.7

However, such a survey may be unnecessary as, in January 2009, the SEC adopted rules requiring public companies to furnish financial statements in eXtensible Business Reporting Language (XBRL). XBRL was developed in the late 1990s under the auspices of what became XBRL International. The US effort was led by XBRL US, which worked with the Securities and Exchange Commission to develop GAAP taxonomies. While the SEC launched a voluntary XBRL filing program in 2005, only a small number of firms participated and data quality was often poor. By mid-2011, essentially all public companies were filing XBRL exhibits with their 10-Ks and 10-Qs. In 2018, the SEC adopted rules requiring Inline XBRL (iXBRL) in which human-readable filings and XBRL tags are combined and today essentially all SEC filers use Inline XBRL.

The SEC publishes two bulk XBRL data libraries that are useful for financial statement research: Financial Statement and Financial Statement and Notes. The latter SEC website explains that “the Financial Statement and Notes Data Sets … provide the text and detailed numeric information from all financial statements and their notes … in a flattened format to help users analyze and compare corporate disclosure information over time and across registrants.”

The Financial Statement and Notes data library comprises eight tables:

  • sub includes one record for each XBRL submission. It also includes fields pertinent to the submission and the filing entity.
  • tag contains all standard taxonomy tags (not just those appearing in submissions to date) and all custom taxonomy tags defined in the submissions. The standard tags are derived from taxonomies in the SEC’s standard taxonomies file as of the date of submission.
  • dim contains all of the combinations of XBRL axis and member used to tag any submission.
  • num contains numeric data, one row per data point in the financial statements.
  • txt contains non-numeric data, one row per data point in the financial statements.
  • ren summarizes “rendering” the data provided by filers about each presentation group as defined in EDGAR filer manual.
  • pre contains one row for each line of the financial statements tagged by the filer.
  • cal is the calculation data set and provides information about arithmetic relations between tagged amounts in a filing.

My dera.pq package downloads those zip files and writes their component tables as Parquet files in a local Parquet repository. More information about making a Parquet repository is found in Empirical Research in Accounting: Tools and Methods, which has both R and Python editions.

3 Analysis

3.1 Extracting cash flow data from XBRL data

In this version of the note, I use R to analyse the data. Specifically, I use the Tidyverse packages along with DuckDB. DuckDB allows me to spin up a fast and powerful database process using a single line of code. I access the DERA Notes data using the load_parquet() function from my farr package, which understands the structure of my Parquet data repository.

In my analysis for this note, I primarily use three tables from those described above, namely, sub, pre, and cal. In the discussion below, I append a _notes suffix to make it clear that I am using the DERA Notes (dera_notes) version of these tables. I also use num_notes for examples focused on Costco’s 2025 10-K.

3.2 Submissions data

The sub_notes table contains data on submissions, including the firm identifier (cik), the SIC code (sic), the period for the filing (period), the type of filings (form) and the timestamp on which the filing was accepted (accepted). In this note, I focus exclusively on 10-K filings (10-K), ignoring amended 10-K filings (10-K/A), filings by foreign issuers (e.g., 20-F) and transition filings (10-KT). While Table 1 shows that there are other annual report forms on sub_notes, 10-K filings are the vast majority of filings.8

Table 1: Annual report forms on sub_notes
form Description Number of filings
10-K Annual report 94,631
20-F Foreign private issuers (non-Canadian, US listed) 9,239
10-K/A Amended 10-K (restated financials, corrections) 8,381
40-F Qualifying Canadian companies 1,346
20-F/A Amended 20-F 1,176
10-KT Covers short periods when changing fiscal year-ends 316
40-F/A Amended 40-F 132
10-KT/A Amended 10-KT 42

3.3 The case of Costco’s 2025 10-K filing

The accession number for Costco’s 2025 10-K filing is 0000909832-25-000101. Table 2 shows some selected data from sub_notes for this filing.9 Looking at accepted in Table 2 causes me to wonder if a filing after 9pm EDT is normal for Costco.10 As it is quite trivial to address this question using the dera.pq data, I create Table 3. From Table 3, it does seem that most filings are after standard business hours for the US East Coast (Costco is headquartered in the US/Pacific time zone), but rarely as late as 9pm EDT.

Table 3 also includes lag, which is the time (in days) between period and accepted. Interestingly, there’s no clear pattern to lag over time, though it does appear to be trending downwards and to vary from one year to the next.

Table 2: Selected sub_notes data for Costco’s 2025 10-K filing
cik name period form accepted
909832 COSTCO WHOLESALE CORP /NEW 2025-08-31 10-K 2025-10-07 21:16:00
Table 3: Filing times and lags for Costco’s 10-K filings
cik period form accepted lag
909832 2025-08-31 10-K 2025-10-07 21:16:00 37
909832 2024-08-31 10-K 2024-10-08 18:40:00 38
909832 2024-08-31 10-K 2024-10-08 18:40:00 38
909832 2023-08-31 10-K 2023-10-10 19:13:00 40
909832 2022-08-31 10-K 2022-10-04 21:59:00 34
909832 2021-08-31 10-K 2021-10-05 20:49:00 35
909832 2020-08-31 10-K 2020-10-06 19:08:00 36
909832 2019-08-31 10-K 2019-10-10 18:06:00 40
909832 2018-08-31 10-K 2018-10-26 09:22:00 56
909832 2017-08-31 10-K 2017-10-17 21:18:00 47
909832 2016-08-31 10-K 2016-10-11 19:26:00 41
909832 2015-08-31 10-K 2015-10-13 18:50:00 43
909832 2014-08-31 10-K 2014-10-15 17:30:00 45
909832 2013-08-31 10-K 2013-10-16 15:40:00 46
909832 2012-08-31 10-K 2012-10-19 17:26:00 49
909832 2011-08-31 10-K 2011-10-14 17:04:00 44
909832 2010-08-31 10-K 2010-10-15 18:41:00 45

3.3.1 Presentation data

I now turn to the pre_notes table, which provides information about the presentation of XBRL data in the associated SEC filing. The variable stmt describes the statement to which each row of the data relates. Table 4 provides information about the distribution of this variable across all 10-K filings on the DERA Notes data.

As might be expected, most data items are not associated with any statement (i.e., stmt is NA), presumably because they are contained in the footnotes to the financial statements. Comparing Table 4 with Table 1 confirms that the vast majority of 10-K filings have all four core financial statements (i.e., BS, CF, IS, and EQ) on pre_notes. Further research would be required to understand whether the cases without one or more of these financial statements can be attributed to the absence of these statements in the filings, absence from the XBRL data supplied with the filing, or some failure in the extraction of the data into the pre_notes data supplied with DERA Notes.

Table 4: Distribution of stmt for 10-K filings
stmt Statement Number of filings Total rows Rows per filing
NA NA 94,477 59,961,029 634.66
BS Balance Sheet 94,160 4,146,278 44.03
CF Cash Flow 93,284 3,978,513 42.65
EQ Equity 92,505 2,544,688 27.51
IS Income Statement 93,318 2,448,181 26.23
CP Cover Page 90,690 2,187,906 24.13
CI Comprehensive Income 41,596 409,983 9.86
UN Unclassifiable Statement 4,607 245,670 53.33
SI Schedule of Investments 878 66,184 75.38

The fields I use from pre_notes are adsh, line, tag, plabel, prole, and negating. To better understand the content of pre_notes, I present a sample of the underlying data in Table 5 ordered by line (for reasons of space, I omit adsh; it is always 0000909832-25-000101 here). As can be seen there, each line is associated with a tag and negating will be used to handle the values I get from cal_notes in later tables.

For my analysis here, the only use I make of prole is to filter out the difficult-to-handle row where prole equals “periodStartLabel”.11

Due to space constraints, I present plabel values in a separate table Table 6. The SEC DERA documentation describes plabel as “the text presented on the line item, also known as a ‘preferred’ label”. That plabel is a little more human-readable can be seen in Table 6.

Table 5: Data from pre_notes for Costco 2025 cash flows
line tag prole negating
1 NetCashProvidedByUsedInOperatingActivitiesAbstract terseLabel FALSE
2 NetIncomeLoss terseLabel FALSE
3 AdjustmentsToReconcileNetIncomeLossToCashProvidedByUsedInOperatingActivitiesAbst terseLabel FALSE
4 DepreciationDepletionAndAmortization terseLabel FALSE
5 OperatingandFinancingLeaseRightofUseAssetAmortization terseLabel FALSE
6 ShareBasedCompensation terseLabel FALSE
7 ImpairmentOfAssetsAndOtherNonCashOperatingActivitiesNet terseLabel FALSE
8 IncreaseDecreaseInOperatingCapitalAbstract terseLabel FALSE
9 IncreaseDecreaseInInventories negatedLabel TRUE
10 IncreaseDecreaseInAccountsPayable terseLabel FALSE
11 IncreaseDecreaseInOtherOperatingCapitalNet negatedLabel TRUE
12 NetCashProvidedByUsedInOperatingActivities totalLabel FALSE
13 NetCashProvidedByUsedInInvestingActivitiesAbstract terseLabel FALSE
14 PaymentsToAcquirePropertyPlantAndEquipment negatedLabel TRUE
15 PaymentsToAcquireShortTermInvestments negatedLabel TRUE
16 ProceedsFromSaleAndMaturityOfMarketableSecurities terseLabel FALSE
17 PaymentsForProceedsFromOtherInvestingActivities negatedLabel TRUE
18 NetCashProvidedByUsedInInvestingActivities totalLabel FALSE
19 NetCashProvidedByUsedInFinancingActivitiesAbstract terseLabel FALSE
20 RepaymentsOfShortTermDebt negatedTerseLabel TRUE
21 ProceedsFromShortTermDebt terseLabel FALSE
22 RepaymentsOfLongTermDebt negatedLabel TRUE
23 ProceedsFromIssuanceOfLongTermDebt terseLabel FALSE
24 PaymentsRelatedToTaxWithholdingForShareBasedCompensation negatedLabel TRUE
25 PaymentsForRepurchaseOfCommonStock negatedLabel TRUE
26 PaymentsOfDividendsCommonStock negatedTerseLabel TRUE
27 FinancingLeasePaymentsAndOtherFinancingActivitiesNet negatedTerseLabel TRUE
28 NetCashProvidedByUsedInFinancingActivities totalLabel FALSE
29 EffectOfExchangeRateOnCashCashEquivalentsRestrictedCashAndRestrictedCashEquivale terseLabel FALSE
30 CashCashEquivalentsRestrictedCashAndRestrictedCashEquivalentsPeriodIncreaseDecre totalLabel FALSE
31 CashCashEquivalentsRestrictedCashAndRestrictedCashEquivalents periodStartLabel FALSE
32 CashCashEquivalentsRestrictedCashAndRestrictedCashEquivalents periodEndLabel FALSE
33 SupplementalCashFlowInformationAbstract terseLabel FALSE
34 InterestPaidNet terseLabel FALSE
35 IncomeTaxesPaid terseLabel FALSE
36 CashFlowNoncashInvestingAndFinancingActivitiesDisclosureAbstract terseLabel FALSE
37 DividendsPayableCurrent terseLabel FALSE
38 CapitalExpendituresIncurredButNotYetPaid terseLabel FALSE
Table 6: Label values from pre_notes for Costco 2025 operating cash flows
line plabel
1 CASH FLOWS FROM OPERATING ACTIVITIES
2 NET INCOME
3 Adjustments to reconcile net income including noncontrolling interests to net cash provided by operating activities:
4 Depreciation and amortization
5 Non-cash lease expense
6 Stock-based compensation
7 Impairment of assets and other non-cash operating activities, net
8 Changes in operating assets and liabilities:
9 Merchandise inventories
10 Accounts payable
11 Other operating assets and liabilities, net
12 Net cash provided by operating activities
13 CASH FLOWS FROM INVESTING ACTIVITIES
14 Additions to property and equipment
15 Purchases of short-term investments
16 Maturities of short-term investments
17 Other investing activities, net
18 Net cash used in investing activities
19 CASH FLOWS FROM FINANCING ACTIVITIES
20 Repayments of short-term borrowings
21 Proceeds from short-term borrowings
22 Repayments of long-term debt
23 Proceeds from issuance of long-term debt
24 Tax withholdings on stock-based awards
25 Repurchases of common stock
26 Cash dividend payments
27 Financing lease payments and other financing activities, net
28 Net cash used in financing activities
29 EFFECT OF EXCHANGE RATE CHANGES ON CASH AND CASH EQUIVALENTS
30 Net change in cash and cash equivalents
31 CASH AND CASH EQUIVALENTS BEGINNING OF YEAR
32 CASH AND CASH EQUIVALENTS END OF YEAR
33 SUPPLEMENTAL DISCLOSURE OF CASH FLOW INFORMATION:
34 Interest
35 Income taxes paid, net
36 SUPPLEMENTAL DISCLOSURE OF NON-CASH ACTIVITIES:
37 Cash dividend declared, but not yet paid
38 Capital expenditures included in liabilities

3.3.2 Numeric data

The num_notes table contains numeric data, one row per data point in the financial statements and the notes thereto. Table 7 provides a sample of such data related to Costco’s 2025 statement of cash flows. Note that the values in uom (the unit of measure) do not vary within this Costco data set. However, qtrs does vary: while most items on Costco’s statement of cash flows have qtrs equal to 4, the value of qtrs for beginning and ending balances of cash and cash equivalents equals 0, as these are point-in-time balances.12

Note that I filter for iprx == 0, where iprx is “a positive integer to distinguish different reported facts that otherwise would have the same primary key. For most purposes, data with iprx greater than 1 are not needed.” I also filter for dimn == 0, as values with dimn greater than zero in this context represent segment data.

Table 7 also includes data on datp. According to the DERA Notes documentation datp represents “the difference between the reported fact date and the month-end rounded date (ddate), expressed as a fraction of 1. For example, a fact reported for 29/Dec, with ddate rounded to 31/Dec, has a datp value of minus 2/31 = -0.0645.” This provides a nice illustration of the reality that, pace the views of RTFM proponents, one should not always trust what one Reads in The Flaming Manual. If you look at the 10-K for Costco, you will see references to “August 31, 2025” (i.e., 0 days before 31 August), “September 1, 2024” (-1 days before 31 August), and “September 3, 2023” (-3 days before 31 August) and these values line up with datp representing the number of days between the fiscal year-end and the month-end (31 August). The field tdate in Table 7 is not actually a field in num_notes, but is a column I calculate using ddate and datp.

Finally, the column dcml indicates the decimal precision of the value column, with -6 meaning the data are reported in millions of the units in uom (i.e., USD).

Table 7: Sample of data from num_notes for Costco 2025 cash flows
tag ddate qtrs uom value datp tdate dcml
IncomeTaxesPaid 2023-08-31 4 USD 2,234,000,000 −3 2023-09-03 −6
IncomeTaxesPaid 2024-08-31 4 USD 2,319,000,000 −1 2024-09-01 −6
IncomeTaxesPaid 2025-08-31 4 USD 2,917,000,000 0 2025-08-31 −6
InterestPaidNet 2023-08-31 4 USD 125,000,000 −3 2023-09-03 −6
InterestPaidNet 2024-08-31 4 USD 129,000,000 −1 2024-09-01 −6
InterestPaidNet 2025-08-31 4 USD 106,000,000 0 2025-08-31 −6
NetIncomeLoss 2023-08-31 4 USD 6,292,000,000 −3 2023-09-03 −6
NetIncomeLoss 2024-08-31 4 USD 7,367,000,000 −1 2024-09-01 −6
NetIncomeLoss 2025-08-31 4 USD 8,099,000,000 0 2025-08-31 −6

3.3.3 Merged data

Now that we understand the data in pre_notes and num_notes, we can merge the data for Costco’s 2025 10-K to produce Tables 8–11. I merge data from pre_notes with data from cal_notes using adsh and tag as the linking variables.13

I present the data from Costco’s 2025 statement of cash flows over several tables partly for reasons of space, starting with cash flows from operating activities in Table 8. I scale the data by dividing the values by one million and I multiply any value where negating is true by minus one.

From Table 7, it can be seen that the num_notes data are in a long form with ddate arranged in a single column, while the typical presentation of statements of cash flows will have the dates organized as columns. Therefore I use the pivot_wider() function to reorganize the data for presentation.14

Table 8 presents the first 12 lines of the merged data set, which relate to cash flows from operating activities. Costco’s cash flows from operating activities are larger than net income—due to the add-back of depreciation and amortization—and growing.

Table 9 presents the next six lines of the merged data set, which relate to cash flows from investing activities. The largest item is capital expenditure (“additions to property and equipment”). That capital expenditures are roughly double depreciation and amortization suggests that Costco is, despite its maturity, still a growing firm.

For reasons I would need to explain in a different note, for standard financial analysis purposes, cash flows would generally be categorized as either operating or financing, with investing cash flows for GAAP purposes relating either to operating activities (e.g., capital expenditures) or to financing activities (e.g., the purchase or sale of short-term investment securities).

A rough-and-ready measure of free cash flows can be calculated by subtracting capital expenditures from cash flows from operating activities. I say rough-and-ready because a more refined measure would do a better job of separating financing cash flows from operating cash flows. For example, it might be appropriate to view the tax withholding on stock-based awards or some portion of the financing lease payments—see Table 10—as operating cash flows despite their classification as financing cash flows in US GAAP.15 It is also probably appropriate to view the interest paid—see Table 11—as financing, not operating, as it is considered in US GAAP. Similarly, some portion of the income taxes paid could be allocated to financing (e.g., the tax savings on interest).

In the case of Costco, the small amount of interest paid means that the rough-and-ready measure will be quite close to a more refined measure.

Table 8: Costco 2025 operating cash flows from pre_notes
2025-08-31 2024-08-31 2023-08-31
CASH FLOWS FROM OPERATING ACTIVITIES


NET INCOME 8,099 7,367 6,292
Adjustments to reconcile net income including noncontrolling interests to net cash provided by operating activities:


Depreciation and amortization 2,426 2,237 2,077
Non-cash lease expense 303 315 412
Stock-based compensation 860 818 774
Impairment of assets and other non-cash operating activities, net 117 9 (495)
Changes in operating assets and liabilities:


Merchandise inventories 559 (2,068) 1,228
Accounts payable 404 1,938 (382)
Other operating assets and liabilities, net 801 741 172
Net cash provided by operating activities 13,335 11,339 11,068
Table 9: Costco 2025 investing cash flows from pre_notes
2025-08-31 2024-08-31 2023-08-31
CASH FLOWS FROM INVESTING ACTIVITIES


Additions to property and equipment (5,498) (4,710) (4,323)
Purchases of short-term investments (1,028) (1,470) (1,622)
Maturities of short-term investments 1,141 1,790 937
Other investing activities, net 74 (19) 36
Net cash used in investing activities (5,311) (4,409) (4,972)
Table 10: Costco 2025 finance cash flows from pre_notes
2025-08-31 2024-08-31 2023-08-31
CASH FLOWS FROM FINANCING ACTIVITIES


Repayments of short-term borrowings (862) (920) (935)
Proceeds from short-term borrowings 816 928 917
Repayments of long-term debt (103) (1,077) (75)
Proceeds from issuance of long-term debt 0 498 0
Tax withholdings on stock-based awards (393) (315) (303)
Repurchases of common stock (903) (700) (676)
Cash dividend payments (2,183) (9,041) (1,251)
Financing lease payments and other financing activities, net (147) (137) (291)
Net cash used in financing activities (3,775) (10,764) (2,614)
Table 11: Remaining Costco 2025 cash flow data from pre_notes
2025-08-31 2024-08-31 2023-08-31 2022-08-31
EFFECT OF EXCHANGE RATE CHANGES ON CASH AND CASH EQUIVALENTS 6 40 15
Net change in cash and cash equivalents 4,255 (3,794) 3,497
CASH AND CASH EQUIVALENTS END OF YEAR 14,161 9,906 13,700 10,203
SUPPLEMENTAL DISCLOSURE OF CASH FLOW INFORMATION:



Interest 106 129 125
Income taxes paid, net 2,917 2,319 2,234
SUPPLEMENTAL DISCLOSURE OF NON-CASH ACTIVITIES:



Cash dividend declared, but not yet paid 0 0 452
Capital expenditures included in liabilities 193 203 170

3.3.4 Calculation data

Finally, I turn to the data in cal_notes, which has the fields adsh, grp, arc, negative, ptag, pversion, ctag, and cversion. Table 12 provides a sample of observations from that table for Costco’s 10-K filing. It turns out that rows with grp equal to 7 relate to the income statement, those with grp equal to 9 relate to the balance sheet and those with grp equal to 14 relate to the statement of comprehensive income. The rows related to the statement of cash flows have grp equal to 3. These rows are shown in Table 13.

Table 12: Data on cal_notes for Costco’s 2025 10-K
grp arc ptag ctag cversion
7 1 NetIncomeLoss IncomeLossAttributableToParent us-gaap/2025
7 2 IncomeLossAttributableToParent OperatingIncomeLoss us-gaap/2025
7 3 OperatingIncomeLoss RevenueFromContractWithCustomerExcludingAsses us-gaap/2025
7 4 OperatingIncomeLoss CostOfGoodsAndServicesSold us-gaap/2025
7 5 OperatingIncomeLoss SellingGeneralAndAdministrativeExpense us-gaap/2025
7 6 IncomeLossAttributableToParent InterestExpense us-gaap/2025
7 7 IncomeLossAttributableToParent InterestAndOtherIncome us-gaap/2025
7 8 NetIncomeLoss IncomeTaxExpenseBenefit us-gaap/2025
9 1 Assets AssetsCurrent us-gaap/2025
9 2 AssetsCurrent CashAndCashEquivalentsAtCarryingValue us-gaap/2025
9 3 AssetsCurrent ShortTermInvestments us-gaap/2025
9 4 AssetsCurrent ReceivablesNetCurrent us-gaap/2025
9 5 AssetsCurrent InventoryNet us-gaap/2025
9 6 AssetsCurrent OtherAssetsCurrent us-gaap/2025
9 7 Assets PropertyPlantAndEquipmentNet us-gaap/2025
9 8 Assets OtherAssetsNoncurrent us-gaap/2025
9 9 Assets OperatingLeaseRightOfUseAsset us-gaap/2025
9 10 LiabilitiesAndStockholdersEquity Liabilities us-gaap/2025
9 11 Liabilities LiabilitiesCurrent us-gaap/2025
9 12 LiabilitiesCurrent AccountsPayableCurrent us-gaap/2025
9 13 LiabilitiesCurrent EmployeeRelatedLiabilitiesCurrent us-gaap/2025
9 14 LiabilitiesCurrent AccruedLiabilitiesCurrent us-gaap/2025
9 15 LiabilitiesCurrent DeferredRevenueCurrent us-gaap/2025
9 16 LiabilitiesCurrent OtherLiabilitiesCurrent us-gaap/2025
9 17 Liabilities LongTermDebtNoncurrent us-gaap/2025
9 18 Liabilities OperatingLeaseLiabilityNoncurrent us-gaap/2025
9 19 Liabilities OtherLiabilitiesNoncurrent us-gaap/2025
9 20 LiabilitiesAndStockholdersEquity StockholdersEquity us-gaap/2025
9 21 StockholdersEquity PreferredStockValue us-gaap/2025
9 22 StockholdersEquity CommonStockValue us-gaap/2025
9 23 StockholdersEquity AdditionalPaidInCapitalCommonStock us-gaap/2025
9 24 StockholdersEquity AccumulatedOtherComprehensiveIncomeLossNetOfT us-gaap/2025
9 25 StockholdersEquity RetainedEarningsAccumulatedDeficit us-gaap/2025
14 1 ComprehensiveIncomeNetOfTax NetIncomeLoss us-gaap/2025
14 2 ComprehensiveIncomeNetOfTax OtherComprehensiveIncomeLossForeignCurrencyTr us-gaap/2025
Table 13: Data on cal_notes for Costco’s 2025 10-K
arc ptag ctag
1 CashCashEquivalentsRestrictedCashAndRestricte NetCashProvidedByUsedInOperatingActivities
2 NetCashProvidedByUsedInOperatingActivities IncreaseDecreaseInAccountsPayable
3 NetCashProvidedByUsedInOperatingActivities NetIncomeLoss
4 NetCashProvidedByUsedInOperatingActivities IncreaseDecreaseInInventories
5 NetCashProvidedByUsedInOperatingActivities DepreciationDepletionAndAmortization
6 NetCashProvidedByUsedInOperatingActivities ShareBasedCompensation
7 NetCashProvidedByUsedInOperatingActivities ImpairmentOfAssetsAndOtherNonCashOperatingAct
8 NetCashProvidedByUsedInOperatingActivities IncreaseDecreaseInOtherOperatingCapitalNet
9 NetCashProvidedByUsedInOperatingActivities OperatingandFinancingLeaseRightofUseAssetAmor
10 CashCashEquivalentsRestrictedCashAndRestricte NetCashProvidedByUsedInInvestingActivities
11 NetCashProvidedByUsedInInvestingActivities PaymentsToAcquireShortTermInvestments
12 NetCashProvidedByUsedInInvestingActivities ProceedsFromSaleAndMaturityOfMarketableSecuri
13 NetCashProvidedByUsedInInvestingActivities PaymentsToAcquirePropertyPlantAndEquipment
14 NetCashProvidedByUsedInInvestingActivities PaymentsForProceedsFromOtherInvestingActiviti
15 CashCashEquivalentsRestrictedCashAndRestricte NetCashProvidedByUsedInFinancingActivities
16 NetCashProvidedByUsedInFinancingActivities RepaymentsOfShortTermDebt
17 NetCashProvidedByUsedInFinancingActivities ProceedsFromShortTermDebt
18 NetCashProvidedByUsedInFinancingActivities RepaymentsOfLongTermDebt
19 NetCashProvidedByUsedInFinancingActivities ProceedsFromIssuanceOfLongTermDebt
20 NetCashProvidedByUsedInFinancingActivities PaymentsRelatedToTaxWithholdingForShareBasedC
21 NetCashProvidedByUsedInFinancingActivities PaymentsForRepurchaseOfCommonStock
22 NetCashProvidedByUsedInFinancingActivities PaymentsOfDividendsCommonStock
23 NetCashProvidedByUsedInFinancingActivities FinancingLeasePaymentsAndOtherFinancingActivi
24 CashCashEquivalentsRestrictedCashAndRestricte EffectOfExchangeRateOnCashCashEquivalentsRest
Figure 1: Calculation hierarchy for Costco cash flow statements

From Table 12 and Table 13, it should be clear that the data in cal_notes describe a hierarchy of calculations, with ctag representing an item that is a “child” of the item represented by ptag (i.e., “parent tag”). I represent the hierarchy embedded in Table 13 in Figure 1.

In some ways, the calculation hierarchy in Figure 1 resembles a typical statement of cash flows, with an overall change in cash sitting at the root of the tree as the parent of the tags representing cash flows from operating activities, investing activities, financing activities and the effect of exchange rate changes. This second level of the tree then has child nodes representing more detailed cash flow items. Each item can be aggregated up the tree to produce the cash flow totals at the higher levels, with due note being taken of negative in doing so.

Looking at Figure 1, we can see also that Costco uses three custom tags in its cash flow calculations. In each case, it seems Costco has combined multiple standard taxonomy items into a single line. FinancingLeasePaymentsAndOtherFinancingActivitiesNet seems to combine FinanceLeasePrincipalPayments, which covers the principal portion of finance lease payments, and “other” financing cash flows.16 ImpairmentOfAssetsAndOtherNonCashOperatingActivitiesNet combines AssetImpairmentCharges and some catch-all item like OtherNoncashIncomeExpense. OperatingandFinancingLeaseRightofUseAssetAmortization presumably combines amortization of both operating and financing leases.17

While it seems clear that Costco has used custom tags to combine multiple standard taxonomy items to align with its chosen presentation in the statement of cash flows, the use of custom tags does have the unfortunate effect of making it more difficult to compare the cash flow data across firms, as other firms may not make the same presentation choices and therefore may use standard US GAAP tags or use different custom tags. In cases like this, perhaps the better approach would be to require the use of standard tags at a lower level but allow firms to aggregate amounts using custom tags purely for presentation purposes.

3.4 Population analysis

Now that we have a better understanding of the data from examination of Costco’s statements of cash flows, I move on to an analysis of the broader population of 10-K filings on DERA Notes.

3.4.1 Calculation data

Recall that my goal is the classification of the operating portions of statements of cash flows based on tags used for operating cash flows in the XBRL data available in DERA Notes. For filings like that provided by Costco’s 2025 10-K filing, it seems that using the data in cal_notes will provide the most reliable approach. In effect, all I have to do is to start at the top of the hierarchy of calculation relationships and then work down to identify the tag that represents cash flows from operating activities and then to identify the tags that are children of that tag in the calculation hierarchy.

Once I have identified the tags representing cash flows from operating activities, it should be fairly straightforward to link specific tags to either the direct-method presentation or the indirect-method presentation.18

For this reason, my approach to classifying the presentation method of each filing prioritizes cal_notes data where available and reliable.

Unfortunately, not all filings have cal_notes data and, of those that do, not all have a cal structure that is regular enough to allow for the reliable identification of the operating-section leaf-node tags.

Table 14 provides periods from 2009 through 2025 and filings on Form 10-K. On the basis of Table 14, I focus on periods between 2014 and 2025, because earlier filings do not reliably include cal_notes data and 2026 is incomplete at the time of writing.19

Table 15 provides information on the regularity of the structure for filings with available cal data. Using recursive common table expressions (CTEs), I create trees for each filing on cal_notes and classify the structure of the cal_notes data for each filing as either regular or irregular, with regularity defined in terms of the presence of an operating cash flow node alongside nodes for investing and financing cash flows.

For filings classified as regular in Table 15, I use operating-section leaf-node tags identified through the cal_notes linkbase to determine the reporting method, as shown in Table 16.

Table 14: Presence of cal data in 10-K filings
year No cal data On cal Total filings Percent on cal
2009 395 9 404 2.23%
2010 1,221 8 1,229 0.65%
2011 6,074 39 6,113 0.64%
2012 6,661 339 7,000 4.84%
2013 621 6,336 6,957 91.07%
2014 27 6,814 6,841 99.61%
2015 16 6,519 6,535 99.76%
2016 11 6,175 6,186 99.82%
2017 6 5,927 5,933 99.90%
2018 6 5,781 5,787 99.90%
2019 15 5,619 5,634 99.73%
2020 8 5,779 5,787 99.86%
2021 21 6,477 6,498 99.68%
2022 61 6,255 6,316 99.03%
2023 24 5,823 5,847 99.59%
2024 118 5,468 5,586 97.89%
2025 26 5,467 5,493 99.53%
Table 15: Regular and irregular cal observations in 10-K filings
Year On cal Regular1 Irregular2 Percent regular
2014 6,814 5,629 1,185 82.61%
2015 6,519 5,484 1,035 84.12%
2016 6,175 5,234 941 84.76%
2017 5,927 5,133 794 86.60%
2018 5,781 5,058 723 87.49%
2019 5,619 4,920 699 87.56%
2020 5,779 5,160 619 89.29%
2021 6,477 5,872 605 90.66%
2022 6,255 5,667 588 90.60%
2023 5,823 5,239 584 89.97%
2024 5,468 4,858 610 88.84%
2025 5,467 4,822 645 88.20%
1 A filing is regular if its cal linkbase has operating, investing, and financing sections under a period-increase-decrease root, or all three section subtotals appear as top-level roots.
2 A filing is irregular if the cal linkbase does not yield a complete set of operating, investing, and financing sections.
Table 16: Direct- and indirect-method regular filers by year
direct1 indirect Total Percent direct
2014 21 5,608 5,629 0.37%
2015 15 5,469 5,484 0.27%
2016 16 5,218 5,234 0.31%
2017 21 5,112 5,133 0.41%
2018 17 5,041 5,058 0.34%
2019 17 4,903 4,920 0.35%
2020 13 5,147 5,160 0.25%
2021 10 5,862 5,872 0.17%
2022 14 5,653 5,667 0.25%
2023 10 5,229 5,239 0.19%
2024 9 4,849 4,858 0.19%
2025 9 4,813 4,822 0.19%
Total 172 62,904 63,076 0.27%
1 A filing is classified as direct-method if at least 80% of classifiable operating-activity leaf-node tags in the cal linkbase are direct-method XBRL tags.
Regular filings are those where the cal linkbase identifies operating, investing, and financing sections with a standard structure (either under a period-increase-decrease root, or as top-level roots). Method classification uses cal_notes.
Table 17: Direct- and indirect-method filers without regular cal data
No cal data Irregular Total Unclassifiable1 Direct Indirect Percent direct
2014 27 1,185 1,212 113 19 1,080 1.73%
2015 16 1,035 1,051 88 17 946 1.77%
2016 11 941 952 84 6 862 0.69%
2017 6 794 800 71 6 723 0.82%
2018 6 723 729 93 15 621 2.36%
2019 15 699 714 62 4 648 0.61%
2020 8 619 627 49 4 574 0.69%
2021 21 605 626 58 3 565 0.53%
2022 61 588 649 114 3 532 0.56%
2023 24 584 608 68 1 539 0.19%
2024 124 610 734 68 2 664 0.30%
2025 26 645 671 103 2 566 0.35%
Total 345 9,028 9,373 971 82 8,320 0.98%
1 A filing is unclassifiable if fewer than 80% of its cash flow statement tags on `pre_notes` are recognisable as either direct- or indirect-method tags.
Filings without cal data or with irregular cal structure are classified using all cash flow statement tags on pre_notes. The direct-method threshold is 80% of classifiable tags.
Table 18: Combined direct- and indirect-method classification by year
indirect direct Total Percent direct
2014 6,673 40 6,713 0.60%
2015 6,397 32 6,429 0.50%
2016 6,061 22 6,083 0.36%
2017 5,825 27 5,852 0.46%
2018 5,657 32 5,689 0.56%
2019 5,541 21 5,562 0.38%
2020 5,718 17 5,735 0.30%
2021 6,424 13 6,437 0.20%
2022 6,183 17 6,200 0.27%
2023 5,767 11 5,778 0.19%
2024 5,512 11 5,523 0.20%
2025 5,378 11 5,389 0.20%
Total 71,136 254 71,390 0.36%

3.4.2 Presentation data

For the remaining filings—those lacking cal_notes data or with irregular cal_notes data—I fall back on pre_notes data, applying the same tag-based direct- versus indirect-method classification to all cash flow statement tags on pre_notes. A filing is unclassifiable under this approach if none of its cash flow tags are recognisable as either direct- or indirect-method tags. Table 17 reports the results of this pre_notes-based classification; the “No cal data” and “Irregular” columns match the corresponding entries in Table 14 and Table 15, respectively.

Table 19 shows that almost all 10-K filings have some data on pre_notes. However, a small percentage of 10-K filings have no data for one or more of the key statements: balance sheet (“No BS data”), income statement (“No IS data”), or cash flows (“No CF data”). The “No FS data” indicates the number of cases where there is no data on pre_notes for any of these three key financial statements. Table 19 also suggests that the availability of filings on pre_notes is high from 2011 onwards, apart from a blip in 2013. The 2026 data cover only part of the year, so relatively few firms have filed 10-Ks for periods ending in calendar 2026.

From Table 20, we see that there are 706 filings with no CF data on pre_notes during the period I focus on (2014–2025). All of the cases are either irregular or no-cal filings, so they fall within the 971 unclassifiable cases in Table 17. For these filings, with no usable data from either cal_notes or pre_notes, there is no systematic basis for classification.

Table 19: Presence of stmt for 10-K filings: All years
No pre data No FS data No BS data No IS data No CF data Number of filings
2004 0 0 0 0 0 1
2005 0 0 0 0 0 1
2006 0 0 0 0 0 1
2007 0 0 0 0 0 1
2008 0 0 0 0 0 2
2009 2 2 3 2 2 404
2010 1 1 1 1 1 1,229
2011 2 3 9 6 78 6,113
2012 1 1 11 11 77 7,000
2013 4 130 145 946 387 6,957
2014 1 2 8 15 69 6,841
2015 0 0 3 5 59 6,535
2016 0 0 5 5 58 6,186
2017 0 0 4 6 48 5,933
2018 0 0 5 5 37 5,787
2019 0 7 8 12 42 5,634
2020 0 5 7 11 38 5,787
2021 0 12 14 20 46 6,498
2022 0 63 64 73 103 6,316
2023 0 18 20 24 54 5,847
2024 0 18 19 26 58 5,586
2025 1 47 51 50 94 5,493
2026 0 0 0 1 2 385
Total 12 309 377 1,219 1,253 94,537
Table 20: Presence of stmt for 10-K filings
No pre data No FS data No BS data No IS data No CF data Number of filings
2014 1 2 8 15 69 6,841
2015 0 0 3 5 59 6,535
2016 0 0 5 5 58 6,186
2017 0 0 4 6 48 5,933
2018 0 0 5 5 37 5,787
2019 0 7 8 12 42 5,634
2020 0 5 7 11 38 5,787
2021 0 12 14 20 46 6,498
2022 0 63 64 73 103 6,316
2023 0 18 20 24 54 5,847
2024 0 18 19 26 58 5,586
2025 1 47 51 50 94 5,493
Total 2 172 208 252 706 72,443

3.4.3 Overall classification results

Table 18 combines the data from Table 17 and Table 15 to show the overall distribution of direct- and indirect-method filers by year.

Table 21 and Table 22 identify the most common tags for direct-method and indirect-method statements of cash flows, respectively. These tables confirm that these tags are disjoint sets.

Figure 2 presents trends on direct-method filers over time in a graphical form. Interestingly, the largest group of direct-method filers are in the Fama-French 12-industry group labelled “Money”, which includes banks and insurance companies where cash flow analyses are often considered less useful.

Table 23 identifies those few firms who filed direct-method statements of cash flows for 10-Ks filed for periods ending in 2025.

Table 21: Most common tags: direct-method operating activities
tag n
InterestPaidNet 77
IncomeTaxesPaidNet 76
ProceedsFromInterestReceived 70
PaymentsToSuppliersAndEmployees 67
InterestPaid 66
PaymentsForOtherOperatingActivities 61
ProceedsFromCustomers 51
ProceedsFromInterestAndDividendsReceived 50
PaymentsForOperatingActivities 48
ProceedsFromFeesReceived 41
IncomeTaxesPaid 36
ProceedsFromInsurancePremiumsCollected 25
ProceedsFromSaleOfLoansHeldForSale 25
ProceedsFromOtherOperatingActivities 23
ProceedsFromSaleOfTrustAssetsToPayExpenses 20
PaymentsForOriginationOfMortgageLoansHeldForSale 20
EquityMethodInvestmentDividendsOrDistributions 15
ProceedsFromRoyaltiesReceived 12
CashReceivedForAccruedIncome 12
InsurancePolicyBenefits 12
PaymentsToEmployees 12
PolicyAcquisitionCosts 12
PaymentsForCommissions 11
PaymentsForLossesAndLossAdjustmentExpense 11
Proceedsfromderivativeinstrumentsoperatingactivities 11
Table 22: Most common tags: indirect-method operating activities
tag n
ShareBasedCompensation 47,325
NetIncomeLoss 39,410
IncreaseDecreaseInAccountsReceivable 32,207
ProfitLoss 27,065
IncreaseDecreaseInInventories 26,702
IncreaseDecreaseInPrepaidDeferredExpenseAndOtherAssets 26,461
IncreaseDecreaseInAccountsPayable 25,513
DepreciationDepletionAndAmortization 24,312
DeferredIncomeTaxExpenseBenefit 22,191
IncreaseDecreaseInAccountsPayableAndAccruedLiabilities 21,298
IncreaseDecreaseInOtherOperatingAssets 17,744
IncreaseDecreaseInAccruedLiabilities 15,580
GainsLossesOnExtinguishmentOfDebt 12,815
OtherNoncashIncomeExpense 12,716
Depreciation 12,651
ProvisionForDoubtfulAccounts 12,321
IncreaseDecreaseInOtherOperatingLiabilities 10,737
GainLossOnSaleOfPropertyPlantEquipment 10,334
DepreciationAndAmortization 10,203
IncreaseDecreaseInPrepaidExpense 10,130
AmortizationOfFinancingCosts 9,875
IncreaseDecreaseInDeferredRevenue 8,843
IncreaseDecreaseInOtherOperatingCapitalNet 8,439
IncreaseDecreaseInAccruedIncomeTaxesPayable 8,201
AmortizationOfFinancingCostsAndDiscounts 7,387
Figure 2: Direct-method filers over time by industry group
Table 23: Direct-method filers with periods ending in 2025
Filer SIC code Industry (12) Industry (48)
BIO-RAD LABORATORIES, INC. 3826 BusEq Measuring and Control Equipment
ROYALTY PHARMA PLC 2834 Hlth Pharmaceutical Products
CNO FINANCIAL GROUP, INC. 6321 Money Insurance
CVB FINANCIAL CORP 6022 Money Banking
ERIE INDEMNITY CO 6411 Money Insurance
MAUI LAND & PINEAPPLE CO INC 6500 Money Real Estate
MBIA INC 6351 Money Insurance
MESABI TRUST 6795 Money Trading
REDWOOD MORTGAGE INVESTORS IX 6500 Money Real Estate
CVS HEALTH CORP 5912 Shops Retail
NEXTPLAT CORP 4813 Telcm Communication

References

Harris, T.S., Morsfield, S., 2012. An evaluation of the current state and future of XBRL and interactive data for investors and analysts (White Paper No. 3). Center for Excellence in Accounting; Security Analysis, Columbia Business School.
Mulford, C.W., Comiskey, E.E., 2005. Creative cash flow reporting: Uncovering sustainable financial performance. John Wiley & Sons, Hoboken, NJ.

Footnotes

  1. See Harris and Morsfield (2012) for some quotes of analysts making precisely this point.↩︎

  2. The SEC Python code uses individually downloaded and unzipped text files, limiting easy reproducibility. Second, it uses pandas, a data frame library that struggles with data sets as large as the DERA Notes data used here, even though ~400 GB is not “big data” by any stretch.↩︎

  3. As an example of how limited my use is, I really do nothing with the numeric data in the DERA Notes database.↩︎

  4. I eschew any attempt to enumerate what I think would be needed to make the data more useful, as this seems to stray too far from the limited nature of this note.↩︎

  5. The standard refers to “cash, cash equivalents, and amounts generally described as restricted cash or restricted cash equivalents.”↩︎

  6. Past editions of Accounting Trends and Techniques and its successors can be found at a website maintained by the University of Mississippi.↩︎

  7. This view seems to receive confirmation in the use of outdated statistics by the SEC Chief Accountant in December 2023, as noted above.↩︎

  8. The amended filings potentially introduce issues of double-counting.↩︎

  9. Note that I have displayed the accepted value in the original US/Eastern time zone.↩︎

  10. The US/Eastern time zone is “EDT” for most of the year, with a switch to “EST” occurring shortly after Halloween at the end of October.↩︎

  11. This row relates to the beginning-of-period balance of cash and cash equivalents. The complication is that these rows share the same tag and ddate values as the end-of-period balances and some jiggering of the data would be needed to put the values in the right column. Given that this would be distraction from my task here, I simply omit that row.↩︎

  12. “Cash dividend declared, but not yet paid”, as seen in Table 11, also has qtrs equal to 0, as it’s effectively a balance sheet amount.↩︎

  13. In merging a larger data set, it might make sense to include version as an additional linking variable, but this does not affect the merge in the current case.↩︎

  14. A wrinkle is that the conventional presentation of financial statement information has the more recent periods to the left, but the names_sort = TRUE argument of pivot_wider() will produce data with more recent periods to the right. But using mutate(ddate = fct_rev(factor(ddate))) yields the desired (conventional) presentation.↩︎

  15. There’s some complexity here. For instance, it might make sense to treat the additions to right-of-use assets under leases as involving offsetting constructive cash flows, with an operating outflow offset by a financing inflow.↩︎

  16. For example, PaymentsForProceedsFromOtherFinancingActivities.↩︎

  17. That is, both OperatingLeaseRightOfUseAssetAmortization and FinanceLeaseRightOfUseAssetAmortization.↩︎

  18. See Table 21 and Table 22 for the most common tags for each presentation method.↩︎

  19. Note that this conclusion is based on the DERA Notes data, not the underlying XBRL filings themselves.↩︎