US durable goods: match the rows before reading three signals

01

Revisit September durable goods through three series

Orders rose in the US Census Bureau’s September durable-goods advance report. That direction does not by itself establish the strength of manufacturing demand or how far it will reach supply. This article connects orders and shipments in Table 1 with backlog in Table 2. The starting task is to move beyond the headline and select rows covering the same scope. Figure 1 and the source list identify the reference period and release time. Establish the chosen inputs before advancing an economic interpretation. [C1]

SG Group’s question is whether orders, shipments and backlog can be treated as three separate pieces of evidence for demand. Different names do not ensure different information. If one number is constructed from the other two, counting all three can give the same information repeated weight. Knowing the construction also avoids dismissing an agreement as meaningless. A calculation can distinguish what agreement checks from what still requires other evidence.

This matters when reading semiconductor and equipment-investment news. A broad label such as US durable goods need not measure the same items as the components, machines or suppliers a reader follows. Before placing the amount on a company purchase order beside a published table, identify the products covered by the statistical row and the contracts included in the company record. An industry-wide change and one company’s procurement decision concern different populations.

First, the article recalculates differences using matching published rows. It then uses fictional contracts to examine cancellations, shipment and the amount remaining at month-end. The final worksheet helps readers distinguish comparable parts of their own purchase cohort from parts requiring more documents. It is a reading aid, not a facility for submitting contracts. The required inputs are records covering the same scope and a separate note in which to transcribe and calculate.

02

Choose matching rows rather than the broadest total

Figure 1 selects only the published inputs needed for this reconciliation. Removing row names and retaining just orders and shipments makes both broad totals look ready for subtraction. Matching column labels, however, do not make the row populations match. Their difference may answer another question. Transcribe the classification at the left of the table before calculating. That small step avoids reconstructing scope from memory after copying the numbers.

For this check, use the row for manufacturing with unfilled orders. The broader durable-goods total is useful for observing an aggregate, but it cannot simply replace the row needed to reconcile backlog changes here. This is not a ranking of better and worse series. An aggregate trend calculation and a balance reconciliation require different scopes. Even numbers printed together in one release should be selected for a defined task. [C1]

Tracking lithography production or delivery requires an additional mapping. A statistical product classification, an equipment company’s segment and a buyer’s capital-equipment category can have different boundaries despite similar wording. Assigning an entire company to a statistical row requires evidence about the products and transactions included. Applying a statistical growth rate to a company without that mapping does not provide fresh evidence about that company’s growth.

Complete one comparison row before expanding the table. Record the document, table, row, column, reference month and unit together, then fill the same six fields for the comparison. Do not fill missing definitions merely because names look similar. A company document labelled “equipment,” for example, may concern product orders or an investment budget that also includes construction. Establishing the definition of that row comes before adding more numbers.

For the separate question of production stages and supply, see Separate process capacity from finished supply.

03

Two differences agree when scope is aligned

The inputs in Figure 1 produce the two differences in Figure 3. Orders less shipments for manufacturing with unfilled orders equal the September-minus-August backlog change: both are $13,569 million. The calculation uses the displayed integers without conversion to growth rates or rounding. Keep the original unit for the first subtraction. Mixing a display in billions with a source table in millions makes unit errors harder to distinguish from scope errors.

Subtracting total durable-goods shipments from total orders instead produces $8,616 million. That subtraction is not arithmetically wrong. The problem arises when it is compared with the backlog change above, because the scopes do not match. The $4,953 million difference between the results is not sufficient reason to conclude that the statistics contain an error. First check whether both inputs and the balance used for reconciliation describe the same scope.

Save the formula as well as its result in a monitoring note. A record saying only that both differences were 13,569 would make an accidental row substitution difficult to detect later. Recording the orders row and column, the shipments row and column and both balance months lets another reader identify a changed input. Also distinguish monitoring agreement from monitoring changes in amounts. The former is reconciliation; the latter leads toward economic interpretation.

The bars in Figure 3 do not rank three demand indicators. Two show the aligned reconciliation; the third illustrates a total-based difference unsuitable for that comparison. Unequal lengths should not immediately be read as unequal demand. The graph distinguishes the ability to calculate in the same unit from the ability to compare the same population. To try another month, first rebuild the equivalent of Figure 1 from that month’s source.

04

What the semiconductor footnote can explain

Footnote 4 includes semiconductor shipments and inventories in applicable aggregates while excluding semiconductors from new and unfilled orders. Interest in chips therefore does not justify treating every relevant column as tracking the same products. The footnote is a condition on reading the table, not incidental end matter. Check inclusion for each combination of product group and column. [C1]

Reading that footnote does not establish that the entire residual above is semiconductor shipments. The broad total and the reconciliation row differ in product or industry scope. Assigning a residual to one cause requires components decomposed on the same basis. This article has not obtained such a decomposition. Semiconductor treatment helps explain why totals cannot be substituted without checking; it does not identify every cause of the residual.

Consider a counterexample in which scope differences operate in several directions. One excluded product could widen a gap while another boundary difference narrows it. The remaining total cannot identify each contribution. A small residual need not mean small component differences, and a large residual does not identify a large change in one particular product. Greater numerical precision will not solve that problem. It requires evidence separating the scopes.

To investigate the chip cycle further, map the scope established here to the production or sales evidence needed next. Exclusion from an orders series is not evidence of absent demand, just as inclusion in shipments does not establish future orders. The required evidence differs for equipment, components and finished products. Before filling an unavailable column with another industry’s number, establish which stage of the relevant transaction that evidence measures.

05

Separate cancellations within a purchase ledger

Figure 4 is a fictional seller ledger, not observed statistical data. Assume opening unshipped orders of 50, new contracts of 90, cancellations of 10 and shipments of 70. The monetary unit is arbitrary and represents no actual company, price or currency. Under those assumptions net orders are 80 and closing unshipped orders are 60. Separating additions from subsequent reductions lets readers follow the route into the closing balance.

The M3 definition of new orders is net of cancellations. Do not extend that wording by describing shipments as “net of cancellations” in the same sense. Figure 4 places its cancellation of 10 against new contracts of 90; it does not reconstruct contracts underlying the official statistics. In a business application, establish whether a change record represents cancellation, a quantity change or a price change. A field labelled merely “adjustment” cannot determine its treatment. [C3]

Even the same cancellation of 10 has different implications for new business if it removes a current-period order rather than an older outstanding contract. A balance reconciliation might combine both in one deduction. Studying the ability to win new business, however, requires the cancellation’s contract cohort. Figure 4 is deliberately a minimal balance example, not a model establishing when each contract was originally secured.

Net orders of 80 less shipments of 70 leave 10. Adding the opening 50 produces the closing 60, but does not establish when that 60 will ship. Contracts due next month and contracts due several months later could produce the same total. Connecting the balance to near-term production requires schedules, changes and actual production stages by contract. Tracing the difference is followed by a separate investigation of its time profile.

06

Distinguish constructed agreement from independent confirmation

The Census methodology derives new orders from shipments and the change in unfilled orders, including in the seasonally adjusted series. The reconciliation is therefore not three independently obtained observations converging on the same answer. It checks the relationship between inputs and an output. Counting all three without recognising this construction risks overstating the number of separate pieces of evidence. [C2]

The fictional example in Figure 5 makes the distinction visible. Shipments of 70 and a backlog rising from 50 to 60 produce orders of 80 through the formula. Change shipments to 75 while holding both balances fixed, and the formula returns 85. Calculated orders can change with an input even if a separate order ledger has not changed. This is SG Group’s assumed example for identifying dependencies, not a reconstruction of an official revision.

If a separate contract record covering the same scope establishes net orders of 80, its gap of 5 from the calculated 85 becomes a starting point for investigation. The gap alone cannot determine which is correct. Check contract inclusion, shipment dates and change treatment. An independent record nevertheless adds something important: unlike repeating the original formula, it can test the inputs and their scope from outside that construction.

When combining indicators, draw which information is produced from which inputs rather than looking only at correlation. Giving extra positive weight to shipments and backlog already used in an orders formula concentrates weight on the same change. Conversely, a contract or delivery record with an unrelated scope is not useful merely because it is independent. Useful comparison requires both a different information source and a mapping to the population under investigation.

For a different construction based on survey responses, see Compare how responses become an aggregate.

07

Keep seasonal-adjustment status consistent

All published inputs in this reconciliation have the same seasonally adjusted status. Readers using their own ledgers do not need to make their values converge with these numbers. Transaction records and an adjusted public series are different forms of input. The appropriate records depend on whether the question concerns economic movement beyond recurring monthly patterns or actual shipment amounts for a contract cohort. Do not mix the two tasks in one reconciliation; state the purpose at the top of the note.

Keep the fictional opening 50 and closing 60 but replace shipments of 70 with an adjusted value of 65, and the relationship with the original net orders of 80 breaks. That is a result of changing the processing of only one input, not evidence that the economic outlook has changed. Put adjustment status beside each value. Relying on a filename or column colour can lose that condition when part of a table is copied elsewhere.

Use inputs from the same release version for this calculation. When combining a previously saved August number with September from a newer document, check whether August changed in the newer version. The task does not require collecting every historical data release. It requires aligning the version used for both balances and the intervening transactions. Recording that version helps distinguish a transcription error from a changed input when another calculation differs.

These checks should be a bounded task before economic interpretation. An endless catalogue of versions and adjustments need not bring readers closer to changes in orders. This reconciliation needs the rows, months, unit, adjustment status and construction formula in Figure 1. Once those are established, examine what changes when moving to a company’s contracts. There is no reason to invent unavailable adjusted values to make a business ledger appear statistically precise.

08

Seller shipment and buyer receipt can fall on different dates

Figure 6 assumes one fictional contract crossing a month-end. The seller dispatches on September 29 and records shipment on September 30; the buyer receives the goods on October 2. These are neither actual company delivery dates nor a claim about an official statistical cutoff. They illustrate that the seller and buyer can record different events involving the same goods. Correct records on both sides need not produce matching amounts for the same calendar month.

The M3 shipment definition is not the same as a buyer’s receiving record. That distinction identifies what a procurement team must investigate before comparing warehouse receipts with statistical shipments. Establish whether each document records dispatch, sales recognition, receipt or acceptance. Do not assume that these always occur on the same date or in a fixed order. This news article cannot determine an individual contract’s terms or accounting treatment. [C3]

Define the event that reduces a company purchase backlog and keep it consistent. A ledger increased at order placement and reduced at receipt retains goods in transit at month-end. A ledger reduced on the supplier’s dispatch notice may already remove that same contract. The appropriate basis depends on the business purpose, but the balances are not interchangeable. Do not interpret differences caused by recording events as shortages or cancellations.

A one-month shift will not necessarily solve the mismatch. Transit times differ across contracts, and partial deliveries or changes can defeat a mechanical shift. First establish which movements can be traced through contract numbers or line items, then regroup them into monthly totals. It may be impossible to match a company’s records fully to public statistics. Even so, this sequence helps investigate whether a mismatch concerns timing, products or counterparties.

09

Map one company cohort to the national statistic

Start with a cohort narrower than the whole company: one supplier, one product group, one currency and one period, with contracts and changes available. Internal transfers across sites or orders routed through distributors may otherwise put repeated stages of the same commercial flow into a simple total. Starting narrowly does not reproduce a national statistic in miniature. It makes it possible to explain, from the records, what counts as one transaction and what reduces the balance.

Record whether the comparison is viewed from the buyer or the seller. A buyer’s order may not match the seller’s accepted order because of acceptance timing or conditions. Combining requests, formal contracts and indicative delivery schedules can make planned demand look established. The M3 definition concerns orders supported by binding legal documents, so an internal list of intended purchases is not automatically an equivalent input. [C3]

Matching amounts do not establish matching breadth of demand. Imagine one cohort with 80 from a single counterparty and another with 10 from each of eight. Totals agree, but concentration and the consequences of losing a contract differ. This is not an estimate of company counts in the public statistics. It shows why contract counts and counterparties add information when reading a company total. The national aggregate cannot reveal a company’s transaction distribution.

After completing one cohort, retain a list of excluded contracts. Quietly removing another currency, a change spanning periods or an unclassified component can make the remaining table appear to represent all demand. Exclusion is not inherently wrong, but it limits interpretation. Showing the exclusions also distinguishes a pre-defined comparison from a selection made because the totals happened to agree. Do not move the boundary after seeing which values fit.

For a separate reconciliation of receivables and payments, see Trace receivables and the route of payments.

10

The same closing balance can imply different production paths

Use the fictional closing balance of 60 from Figure 4 to consider two paths. In case A, all 60 is scheduled to ship next month. In case B, 20 is due next month and 40 later. The closing total is identical, but near-term work differs. Connecting backlog growth to faster production requires this time profile. A total alone may suggest work accumulated for the future; it cannot directly determine next month’s production.

Money and workload also differ. Contracts worth 60 could cover one expensive item or many low-value components, with different operations, time requirements and constraints. Converting value into units through a fixed coefficient without quantity detail lets an assumption determine the conclusion. For one company product group, separate contract quantities and prices to investigate changes in value. Having that detail does not establish the national product mix.

In a third fictional case, C, delivery dates remain unconfirmed and part of the 60 is absent from the shipment schedule. That weakens a claim that a large backlog securely supports production. Formal schedules or progress records could reduce that concern. SG Group’s interpretation is neither that backlog growth is always favourable nor that it necessarily signals delay. The assessment changes with evidence about when, and under what conditions, the contracts will be performed.

Documents are more useful than bullish or bearish labels for distinguishing these cases. Balances by due date, quantities and prices, changes and performance stages each address a different uncertainty. If documents are unavailable, do not assign the contract to the most convenient case. A reader relying on public aggregates can likewise identify which missing breakdown limits a conclusion. The limitation makes the next investigation concrete rather than stripping the news of value.

11

Identify transaction location and stage before drawing global implications

A route from US manufacturing statistics to the global economy begins with relationships between buyers and producers. More orders for a supplier may lead it to procure additional parts or equipment. Effects reach production or transport in another country only if the relevant transactions occur and work passes to those counterparties. Aggregate growth alone cannot identify which country’s suppliers benefit. This is a conditional commercial mechanism, not a measurement of contributions to individual countries.

Imagine a US manufacturer accepting an equipment order and buying components abroad. Adding the equipment order, component purchases and finished-product exports as separate new final demand would count stages of the same product repeatedly. Appropriate evidence depends on the stage: operations for manufacturing work, transport or trade records for cross-border goods, and the buyer’s use for final demand. One M3 difference does not measure all those stages on their behalf.

The same restraint applies to markets. Assigning a post-release move in equities, yields or exchange rates to one statistical column requires examining other announcements, prior expectations and trading times. This article has not measured the day’s market reaction. Understanding the orders arithmetic does not establish a price direction. Reading construction correctly can narrow an economic hypothesis; it does not guarantee trading returns or the size of a reaction.

For a particular company’s supply contract, documents linking it to its counterparties provide a more direct path. A national statistic moving in the same direction as company evidence can inform background interpretation, but cannot verify that company’s orders without a scope mapping. Opposite movements do not immediately make either source wrong. Investigate whether geography, products, customers or execution timing separate the company’s experience from the aggregate.

For alignment of a company’s consolidation boundary, see Reconcile assets deducted from borrowings.

12

SG Group View: what changes an interpretation of stronger demand

To evaluate “more orders mean stronger demand,” first specify the period, products and customers meant by demand. Without that boundary, almost any follow-up can be absorbed into the same explanation. A hypothesis about near-term production needs evidence about contracts due soon and shipment progress. A hypothesis about a long-term equipment plan may not change with short-term shipment movements alone. The question’s horizon determines the evidence capable of challenging it.

SG Group places weight on evidence outside the reconciliation formula. More contracts by due date, performance progressing to schedule and cancellations not erasing the increase would support an interpretation of more future work. This does not claim that those conditions were observed in this release. They are conditions for the reader’s investigation. Agreement among orders, shipments and backlog must not be treated as though it also established all of them.

Specify observations that would change the assessment: an increase explained only by repricing, contracts concentrated at distant due dates, later-discovered cancellations, or an expanded scope lifting the aggregate. These are different explanations and should not all be compressed into “weak demand.” A price explanation calls for quantity evidence; a scope expansion calls for a comparable population. Different challenges imply different next tasks.

A reasonable objection is that an advance release cannot reveal detailed contracts, so a broad directional reading is enough. That can be useful for an economic overview. The reconciliation matters when making a stronger claim: that three numbers independently confirm demand, or that a particular supplier’s production will rise. Stronger conclusions require more specific evidence. The usefulness of an advance report and limits on company-level decisions can coexist.

13

Complete one comparison with an eight-field worksheet

Begin Figure 7 by writing one question, such as “Will this month’s increase in orders to this supplier increase units received next month?” Fill in scope, contract formation, changes, shipment, receipt, balances, construction dependencies and additional evidence. If comparing with the public statistic, place its matching row alongside. Readers without contracts can practise using only Figure 4’s assumptions, but should not record that exercise as a completed business verification.

If order-formation timing is unknown, examine the contract and acceptance notice. If quantity and price changes share an adjustment field, return to the line items. If shipment and receipt fall in different months, separate goods in transit. Connecting one mismatch to one document moves the work beyond a vague statement that data are missing. Not every issue must be solved at once, but an unknown field must not enter the calculation as zero or as a completed cancellation.

Distinguish values already present in the ledger from values you calculate. Do not overwrite orders confirmed from contracts with net orders inferred from the balance change. Keeping only the version that fits would remove the discrepancy worth investigating. Put the original record, formula, result and difference alongside one another so later documents show what changed. Record sensitivity trials like Figure 5 separately from actual contract changes.

The output is not an automatic pass or fail or a supplier rating. It is a note explaining which comparisons work, where scope or recording dates differ, and which next document would advance the judgment. Reading and using the table is free, but users collect records, calculate and retain their notes. The site does not obtain company contracts, automatically convert them into statistical equivalents or certify contractual eligibility.

14

At the next release, start by preserving the same reconciliation

At the next release, look beyond prominent percentage changes. First check whether orders less shipments reconcile to the balance change using the same row, reference month and processing status. Record which inputs changed. A fixed procedure discourages switching to whichever series fits the preferred explanation. Another agreement still does not add an independent demand confirmation. The construction relationship remains something to check in the next document.

Advance the economic assessment by separating evidence that supports or changes the hypotheses about timing, contract performance, quantities, prices and scope. Keep findings from a company cohort within that cohort. Extending them to the national aggregate requires a separate mapping. Equally, do not force company records to fit the national number. Preserving both boundaries makes it easier to investigate where an economic change reached actual business activity.

An agreeing table does not eliminate statistical uncertainty. M3 uses a nonprobability sample, so this exercise cannot produce conventional sampling errors or confidence intervals. It reconciles documents rather than independently certifying statistical precision. That limitation does not make an advance release unusable, but it constrains claims that small differences establish certain economic changes. Do not strengthen a conclusion by assuming unmeasured error is zero. [C2]

After reading a summary of this news, a reader can come to SG Group to recalculate the two differences from matching rows and compare one purchase cohort through eight fields. The result can explain whether the same information was counted three times, whether another record added independent evidence, and where contracts remain incomparable. There is no reason to hide the direction of orders. The remaining task is to establish how far that conclusion travels to the reader’s records and choose the evidence needed next.