Why don't we book emissions like money?

In the 13th century the bankers of Florence solved a problem that carbon accounting still has: two parties keeping separate accounts, with neither one's entry pointing to the other's. This article covers what double-entry bookkeeping did, how much of what companies calculate for their purchases is estimated, how accounting researchers and others try to carry the principles over to emissions, how the draft standard MASSIV+ tries to solve what has been identified as difficult, and the obstacles to booking emissions the way we book money.

Two merchants in Florence

Compagno Soldi counts the coins on the table for the third time. It is autumn 1211 in Florence, and in front of him lie six kinds of coin, minted in Pisa, in Lucca and in cities he does not recognise. He has a small workshop where wool is carded, spun and woven into cloth. His surname means money, but money is exactly what he is short of. He buys his wool from Baldovillano di Sotto, a trader with a warehouse a few streets away, and he pays when the cloth is sold. Right now he owes Baldovillano two soldi and six denari, and he wants to pay on time. Without wool the loom stands still.

But counting out the coins is not enough. Coins are scarce in the city, and the ones in circulation come in many kinds, of uncertain value. Baldovillano wants a money changer to weigh and test them first. There are no banknotes.

So, like everyone else, Compagno does most of his business on credit or by barter. A debt needs no coins today. It is written down in lire, soldi and denari and paid later, in coin, in goods or by setting it off against another debt. The lira is not even a coin, just a way of counting: one lira is twenty soldi, one soldo twelve denari. It is the reckoning everyone trusts, not the coins in hand. That is how debts run through the city: Compagno's customers owe him, he owes Baldovillano, and Baldovillano owes his suppliers.

But a debt is hard to prove. Compagno writes the deal down in his own notes, in his own words. Baldovillano writes it in his, and Compagno does not know what they say. If they disagree, there is little to go on. Compagno can give Baldovillano a slip acknowledging the debt, and the slips he has received himself he keeps in a box under the loom. But a slip like that is easy to dispute. If they want to be sure, they have to go to a notary and have him write the deal down, and nobody does that for every small purchase.

When Compagno's customers pay him, he gets coins of every kind. He has to carry them to Baldovillano, at the risk of losing them on the way, and Baldovillano wants them weighed and perhaps exchanged again before he pays his own suppliers. Every link has its own book and its own picture of what has been paid.

At a distance it is harder still. Some of what Compagno needs is bought through an agent at the fairs in Champagne, a long journey north over the Alps. He trusts the agent, but he is not there and does not see what the agent buys or what it costs. He finds out only when the agent writes home after the fair and accounts for what was bought and paid, and the letter is weeks on the road. Meanwhile Compagno works from what such things usually cost.

So Compagno does what more and more merchants in the city are doing and opens an account with a banker. The bankers sit at their tables down by the square, in the shade of the trees between Mercato Vecchio and Mercato Nuovo, and that suits him well, because Baldovillano already has an account with the same banker. Now no coins need to be carried, weighed or exchanged. Baldovillano asks the banker to transfer the amount, and the banker writes two entries: on Compagno's page that he was to give two soldi and six denari, on Baldovillano's that Compagno has given them. Baldovillano can then pay the next person the same way, without anyone touching a coin.

The deal now exists in one place, in a book both of them trust. Compagno is satisfied. But the book is thick, and every customer has his own pages. Compagno's page says that he has given two soldi and six denari to Baldovillano. If he wants to see that the same amount really appears under Baldovillano, he has to leaf through to Baldovillano's pages and search among the entries there. Nothing on his own page tells him where the other entry is, or that the payment really arrived.

When Baldovillano then pays someone else, and whoever receives the money pays on in turn, the result is a chain of entries on different pages of the book. None of the entries points to the next. Anyone who wants to follow the money has to search their way along, entry by entry. In practice nobody does. It is far too laborious.

That Compagno dares to trust the book anyway is down to something he has seen with his own eyes. One morning, as he stands in the queue at the table, two men come walking from the guild's house. The banker puts down his pen and hands over the book without a word. The men leaf through it, point at an entry, ask a question, leaf again. Nobody in the queue says anything. It is the Arte del Cambio, the bankers' own guild, and its inspections come unannounced.

It was the city's merchants who forced the guild into being a few years earlier, so that they could trust the people they left their money with. Compagno was not one of the big men who did the talking, but he was one of the many in the wool trade who, through their own guild, demanded order in the banks. The guild is run by the bankers themselves, not by the merchants. They write the rules for how a bank is to be run, and they inspect one another. Nobody may run a bank in Florence without being a member, so a banker who is expelled is finished. The guild decides how the books are to be kept, with a separate cash book and Roman numerals, which the guild considered harder to forge than the new Arabic ones. The books must be clear, must balance and must be free of fraud, and the tables stand only a few hundred metres from the guild's house, so no banker has time to put anything right before a visit. A banker whose books do not measure up risks a fine or expulsion.

One day Compagno sees the banker write something more beside his entry: a number, after Baldovillano's name. He asks what it is. "The page where his entry is," says the banker, "so the guild does not have to search." That was the whole solution. Beside every entry the bankers now wrote where the matching entry stood, and the inspector could follow every payment from one account to the other and see that they agreed. That is the core of what came to be called double-entry bookkeeping. In return, the banks' books were accepted as evidence when merchants were in dispute, just like a notary's deed. For Compagno it meant that the entry in the bank's book was enough. If he and Baldovillano disagree, that is where they go, not to the notary, and what the banker has written stands. The box under the loom is still there, but the slips in it have become keepsakes. What he needs to know about his dealings is now in a book that someone else inspects, and where every entry points to its counterpart.

The story is largely invented. Compagno Soldi and Baldovillano di Sotto existed, but all we know about them is what is written in one entry in a bank book from 1211: that two soldi and six denari were transferred from one's account to the other's, at Baldovillano's request. The surviving leaves are today in the Laurentian Library in Florence. The bankers were Florentines, but the book was probably kept in Bologna. The workshop, the wool, the warehouse, the agent, the box of slips, Compagno's part in founding the guild, the scene being set in Florence and his seeing the reference being added himself are our invention. Nobody knows when in the 13th century the reference came about. His circumstances are based on the accounting historian Alan Sangster's account of how double-entry bookkeeping probably arose among the bankers of Florence: the lack of reliable coins, trade on credit and through agents at a distance, debts that were hard to prove, payments between accounts, and a guild that prescribed how the books were to be kept and inspected them without warning (Sangster 2016).

Where carbon accounting stands today

That was eight hundred years ago. Even so, anyone who works with carbon accounting will recognise Compagno's situation before he went to the banker, almost point by point.

First a few words for those who do not calculate emissions every day. Most companies calculate according to the Greenhouse Gas Protocol (GHG Protocol), a voluntary standard for how a company reports its emissions, developed by two international organisations, the World Resources Institute and the World Business Council for Sustainable Development. It is the basis of the rules that now require climate reporting: the EU's Corporate Sustainability Reporting Directive (CSRD), the international reporting standard IFRS S2, which more and more countries are adopting, and California's climate disclosure law. The protocol divides a company's emissions into three parts. Scope 1 is what the company itself emits, for example when it burns fuel in its own boilers or runs its own vehicles. Scope 2 is the emissions from the electricity and heat it buys. Scope 3 is the rest of the chain, above all what was emitted when the things the company buys were made. Scope 3 is often the largest part, and the hardest, because the emissions happened somewhere else. To calculate it, emission factors are used: numbers that convert something the company knows, such as kronor, kilos or kilowatt hours, into emissions. A factor can say, for example, how many kilos of carbon dioxide one krona of purchased transport usually corresponds to. Anyone who wants more exact numbers can use an EPD, an environmental product declaration, in which a manufacturer reports a product's environmental impact, including its emissions, calculated per product and verified by a third party.

Few values are actually measured. Instead of coins of many kinds there are emission factors from many databases, with different values for the same purchase, and someone with experience has to judge which one applies, like the money changer at the table. Emissions, like the debts, run between many companies at once. The supplier reports its emissions and the customer calculates its share, and neither sees the other's calculation. What carries the most weight is what a third party has verified: an EPD for a product, or the climate report in an annual report for a company and a year. They are the notary's deeds, one for each product and one for each company. But each is written with its own assumptions, so two EPDs can rarely be compared with each other, and neither can two annual reports. And none of them says what this particular customer got from this particular supplier. When the emissions pass on to the next link they are often recalculated, and what was known at one link becomes an average at the next.

On top of that, anyone who wants to know what the supplier actually emitted has to make do with what it usually is, like Compagno with his agent. The models for doing so are more or less sophisticated, but they all estimate something the supplier itself could know.

The attempts at a solution have their counterparts too. The GHG Protocol plays the guild's role as far as the rules go: it standardises how the calculations are to be done, but every company still keeps its own book. Most carbon accounting platforms provide shared methodological support rather than a shared book. Where product footprints, that is, emissions calculated per product, are actually shared between companies, as in the automotive industry's data collaboration Catena-X, it resembles the banker's book from 1211. The customer fetches the number directly from the supplier, under shared rules for how it is calculated and sent. The number identifies who it comes from and which product it covers, but not where the matching entry stands. Nothing ties what the supplier hands to different customers to its own emissions, and the rulebook itself says it does not guarantee that the footprints agree with companies' Scope 3 accounting (Catena-X 2025, p. 14). The footprints can also be verified by an outside party, who samples meter readings, invoices and the calculation for a product (Catena-X and TfS 2025, p. 33). But the verifier checks one footprint at a time. There is no matching entry to follow a tonne to.

So carbon accounting has the rules and the inspector, and in places shared channels for sending the numbers. What is missing is for every entry to show where the matching entry stands.

If bookkeeping worked like carbon accounting

How much is estimated shows most clearly if the comparison is turned around. The GHG Protocol, the standard most companies calculate by, gives four ways to calculate the emissions from purchases, from the coarsest to the most specific. Imagine money being booked the same way (table 1).

The GHG Protocol's methodWhat is usedIf money were booked that way
Spend-basedAmount times an industry average per kronaThe buyer books the supplier's profit as the amount times the industry's average margin, without reading the invoice
Average-dataQuantity times an industry average per kiloThe buyer books the purchase as the number of kilos times an industry price, whatever the invoice says
HybridThe supplier's own emissions (Scope 1 and 2), the rest of the chain with averagesThe supplier invoices its own work, but everything it bought in itself is priced at industry prices
Supplier-specificThe supplier's EPD or product footprintThe supplier sends a cost estimate of what the product usually costs to make, done a few years ago and with its own assumptions about its purchases

The GHG Protocol's four methods for emissions from purchases, and what each would correspond to if money were booked the same way.

So even the most specific method, with the supplier's EPD or product footprint, is more like a cost estimate than an invoice. It is modelled for one product at one point in time, and two suppliers' cost estimates rest on different assumptions.

The comparison is a little unfair. Suppliers' own numbers are rarely available, and Scope 3 is built so that every company sees its impact across the whole chain, including where it does not own the emissions. The GHG Protocol itself notes that the same emissions therefore end up with several companies (Scope 3 Standard, section 9.6). But the thought experiment shows what emissions reporting lacks.

This is clearest when something changes. When a supplier reduces its emissions, the customer rarely sees it. If the customer uses an industry average per krona, the reduction does not show up at all. If it uses an EPD, the reduction shows up only when the EPD is redone, and then it is hard to tell whether it was the emissions or the assumptions that changed. We have described the factors and how they are chosen separately, in the article on emission factors. The question here is a different one: how could a real reduction show up? The rest of the article is about that: first which principles bookkeeping rests on and how others have tried to carry them over to emissions, then a third proposal.

From the principles of bookkeeping to emissions

What the bankers of Florence arrived at has since been distilled into a few principles. Every transaction is booked by both parties, and the counterparty is named. Debit and credit balance in every entry, and so the totals balance without anyone needing to see the whole economy. Estimates are reported separately. And because everyone follows the same rules, two sets of accounts can be compared.

Carrying those principles over to emissions is not a new idea, and in recent years it has taken shape in accounting research, starting from the same criticism of today's way of calculating. Since 2021 Kaplan and Ramanna have developed what was first called E-liability and is today called E-ledgers, in which every product carries its emissions with it, from the raw material to the factory gate, and passes them on at every purchase, with entries that link the seller's and the buyer's books (Kaplan and Ramanna 2025). Reichelstein builds a balance sheet and an income statement for carbon, modelled on how inventory and fixed assets are accounted for (Reichelstein 2024). Both object that today's numbers rest on industry averages instead of suppliers' actual emissions, and both have each company pass its own emissions on to its customers. Reichelstein states the conclusion plainly: a reduction at the company and its suppliers usually does not show up in the reported footprint (p. 2130). Both begin with industry averages for what suppliers have not yet reported, and expect them to be replaced link by link.

They differ in what is counted, and that is where the difficulties lie.

E-ledgers counts per product. Every company has to allocate its own and its purchased emissions to what it sells, and since different customers buy different products, that allocation has to be right. Kaplan and Ramanna require it to be audited, as thoroughly as a set of financial statements, at every link, because if it is wrong, the number passed on to the customer is wrong too. They point out that in financial reporting it is enough for costs to be allocated to products with simple rules, since the auditor only makes sure the whole adds up, but that this is not enough for emissions (p. 8). The number also does not apply to the product in general but to each batch made, since emissions differ from one run to the next. In the pilot projects, difficulties arose exactly there, with the nearest suppliers "unable or unwilling to provide dynamic, batch-level data for the products they supply" (Kaplan and Ramanna 2025, p. 25).

Reichelstein sidesteps the problem. His main measure is the company's total emissions in what was sold during the year, and that total is largely the same however the company allocates its emissions between its products, as long as inventory neither grows nor shrinks (p. 2138). That is enough to assess the company. But a customer does not buy the whole company, it buys a product, and the number for that particular product depends entirely on how the allocation was made. Reichelstein notes himself that the choice of allocation rule, or key, "will leave companies with discretion in burdening individual products at the expense of others" (Reichelstein 2024, p. 2138). This is a known challenge, and he gives an example in the article where the industries' rulebooks disagree: for the slag from steelmaking, which is used in cement, the steel industry's organisation advocates allocation by weight and the cement industry's by value (p. 2148).

So both proposals run into the same difficulty: a company's emissions have to be divided among the products the company sells, and that division can be made in many ways. Kaplan and Ramanna deal with it by requiring the division to be audited, in every company, for every batch. Reichelstein deals with it by making the total for the whole company the main measure, since the division barely affects it. But the number per product, the one the customer receives, is still affected. Neither asks whether the division has to be made per product at all. MASSIV+, a draft open standard being developed with support from Vinnova within the programme Avancerad digitalisering, takes up exactly that question (docs.massivplus.org).

Divide by customer instead of by product

The basic idea is that a company's own emissions are its customers' purchased emissions. The company measures what it emits itself, adds what its suppliers have passed on, and passes everything on to its customers, in proportion to how much each has bought during the period. The division is therefore made by customer, not by product, and it rests on what the company already knows from its books: whom it has sold to and how much. A number per item arises only at the last link, with whoever sells to a consumer, since the consumer has no customer to pass anything on to. Whoever calculates this way is called a node in the standard.

In the draft from November 2026, four rules correspond to the principles of bookkeeping. The current wording is in the specification.

  1. Emissions are conserved at every node. What the node allocates to its customers adds up to what it has. This corresponds to debit and credit balancing.
  2. Actual or unknown. What a supplier has reported itself is actual. The rest is unknown, even if the customer estimates it. This corresponds to estimates being reported separately.
  3. Replacement. When a supplier starts reporting, its number replaces the estimate, and the customer sees it the next period.
  4. Actual share. Every node shows how much of its emissions rests on reported numbers.

Imagine that Baldovillano in our opening story calculated this way. He would give Compagno one number for everything Compagno bought from him during the period, not one number per sack of wool. Compagno then does the same towards his customers.

So Baldovillano does not have to put a number on every sack. That is a relief for two reasons. To divide by customer he needs only his own book, which he has to keep anyway in order to get paid: who bought and how much. To divide by sack he would need to know which sack came from which farm, went through which wash and travelled on which cart. Nobody keeps books on that, and if all the wool is washed together, there is not even anything to know. Besides, the division by customer is made once per period, as a share of a total, while a number per sack has to be calculated, and be open to audit, for every delivery.

So Compagno gets one number per period from Baldovillano, not one per sack. That is enough for what he can do something about. He can compare Baldovillano with another wool trader and choose the one that emits the least. He can ask Baldovillano to do something about his emissions and see the next period whether it made a difference. And he knows who answers for what: Baldovillano for his number, Compagno for his.

Two things in the description deserve an explanation.

The first is why own and purchased emissions are linked. What is one company's Scope 3 has always been emitted somewhere else: at the supplier, at the supplier's supplier and so on back to the raw material. There it was counted as their Scope 1 and 2. So a company's Scope 1 and 2 becomes its customers' Scope 3. MASSIV+ lets emissions take the same path as the transactions, from supplier to customer. Every node reports its Scope 1 and 2 as in the GHG Protocol, and what it receives is its Scope 3, so the numbers fit straight into an ordinary climate report. What is passed on, on the other hand, is aggregated. The customer sees how much is actual and how much is unknown, but not what was the supplier's fuel, electricity or purchases.

The second is the node. The word is new and chosen deliberately: a node does not have to be a whole company, as in the GHG Protocol, but can be a factory, a department or a line of business, as long as it can say what it emits itself, what it buys and whom it sells to. A company can divide itself into several nodes where needed, for example when part of the business has quite different emissions from the rest, and count as one node where that is enough. All nodes use the same emission factors for their own emissions, and each states whether it allocates by weight or by value.

The proposal lands between the two most common ways of calculating. The GHG Protocol calculates per organisation, but every company in its own way, so two reports are hard to compare. EPDs calculate per product and rest on assumptions about where the calculation should stop and about averages. MASSIV+ calculates per organisation, in smaller parts and by the same rules for everyone. Of the GHG Protocol's four methods for purchases (table 1), the hybrid method comes closest, since it uses the supplier's own emissions. But where the hybrid method fills in the rest of the chain with averages and counts them, MASSIV+ leaves what no supplier has reported as unknown and shows how large a share it is. And that applies at every link, so that the unknown shrinks as suppliers' suppliers start reporting. The idea is that the numbers can then be compared, like two sets of accounts.

Obstacles to booking emissions the way we book money

MASSIV+ borrows its form from financial bookkeeping: one rule for every transaction, booked by both parties. Four obstacles explain why it is not already done. One is a tradition, and it explains why it has taken so long. The other three are points where emissions really are different from money, and they apply to MASSIV+ as well.

  1. The tradition. Carbon accounting inherited its way of thinking from life cycle assessment, the method for calculating a product's environmental impact from raw material to waste. It was developed to compare ways of fulfilling a function: how much the environment is burdened by a litre of milk on the table, or by moving a tonne of goods a thousand kilometres. It calculates per function, and all the materials and all the links that contribute to the function are counted, regardless of who owned them or who sold to whom. When companies began calculating their own emissions in the 2000s, they borrowed life cycle assessment's databases, factors and above all its question: what has this product caused? Bookkeeping has never asked what a coin caused. It asks who gave it to whom. The two questions produce different books, and anyone used to the one does not see the other as an alternative. Changing the question is the biggest step, and it is the step MASSIV+ proposes.

  2. The amount is agreed; the allocation is decided by one party. When Compagno pays Baldovillano, the amount is settled between them. There is nothing to allocate. Baldovillano's emissions, on the other hand, do not belong to any particular sack, and how much of them should fall on Compagno is for Baldovillano himself to decide, by the key he has chosen. Compagno had no part in that choice, and a different key would have given him a different number. The total is the same however Baldovillano divides it, but what each customer gets is not. The standard requires the same key to be applied to all customers during the period, so that nobody can be favoured at another's expense, something bookkeeping has never needed to worry about.

  3. Both parties want a low number. With money they have opposite interests. Baldovillano wants to be paid more, Compagno wants to pay less, and each reads the other's entry with that in mind. In emissions both want a low number, and that is already visible today. A supplier with a green and an ordinary product line can let every customer believe it was the one that got the green, and no customer has reason to ask. The first rule above, that emissions are conserved at every node, is meant to close that loophole. Baldovillano cannot hand out more of something than he has, and Compagno knows what he bought and can see whether his share adds up. But the total being divided, Baldovillano's own emissions, is something he and all his customers want as low as possible, and nobody in the chain checks it. The text of the standard does not yet say who should check it instead. The texts around it suggest that small companies self-certify and large ones are verified by a third party, but precisely where no counterparty recalculates, a self-certification carries less weight than it does with money.

  4. Participation is voluntary. Everyone who does business keeps books, and whoever does not invoice does not get paid. But nobody is obliged to send their emissions number to the customer. A link that does not calculate breaks the chain, and everything further back in the chain from it becomes unknown to everyone who comes after. The standard deals with this by showing how large a share is unknown, instead of filling it in with averages, but it cannot force anyone to take part. Bookkeeping has had the law on its side since the days of the guild, and it applies to everyone who does business. Climate reporting has had the law on its side only in recent years, and it applies only to the largest companies, differs between parts of the world and requires a report on the company itself, not an entry to the customer. Most of Baldovillano's counterparts today are not covered at all.

It also remains to be tested at scale. A single company can start by calculating its own emissions according to the standard, but the effect across a whole chain requires several links to take part.

The obstacles can be overcome. There is an example of how quickly a new arrangement for how companies account between themselves can spread. VAT rests on the same idea as this article: the seller's output VAT is the buyer's input VAT, the entries are linked, and the chain ends with the consumer, who cannot deduct it. Sweden introduced VAT in 1969, and today more than 170 countries have it (OECD, Consumption Tax Trends). It took a few decades, not eight hundred years. But VAT had laws and tax authorities behind it from the start. The bankers of Florence had only a guild and one another's inspection.

About the author

I work at Icons Of, which is one of the parties in MASSIV+ and is responsible for the project's technology and currently for the development of the standard. What is written here is my view, not the project's. So the section on MASSIV+ is written by someone who has a stake in the proposal, not by an outsider who has assessed it.

Sources

  • MASSIV+. Introduktion, teknisk specifikation och fördjupningen Bokföringsanalogin. Draft, 2026. docs.massivplus.org
  • MASSIV+. The project's website. massivplus.org
  • GHG Protocol. Corporate Value Chain (Scope 3) Accounting and Reporting Standard, 2011. Section 9.6. ghgprotocol.org
  • Sangster, A. The Genesis of Double Entry Bookkeeping. The Accounting Review 91(1), 2016, pp. 299-315. doi.org
  • Catena-X. Catena-X Product Carbon Footprint Rulebook, version 4, 2025. catenax-ev.github.io
  • Catena-X and Together for Sustainability. PCF Verification and PCF Program Certification Framework, version 2.0, 2025. catenax-ev.github.io
  • OECD. Consumption Tax Trends. oecd.org
  • Kaplan, R. S. and Ramanna, K. E-ledgers Carbon Accounting. Blavatnik School of Government, working paper BSG-WP-2025-62, 2025. Published in Management Science, 2026. Page references are to the working paper. doi.org
  • Reichelstein, S. Corporate carbon accounting: balance sheets and flow statements. Review of Accounting Studies 29, 2024, pp. 2125-2156. doi.org
  • Icons Of. Which emission factor belongs on each row of the general ledger? 2026. iconsof.se