Chapter 3: Measuring Green Governance: Carbon Accounting, MRV Systems, and the Boundaries of Evidence

In the autumn of 2021, a mid-sized furniture manufacturer in Foshan, Guangdong, received a questionnaire from a European retail buyer. The buyer wanted the carbon footprint of a flat-pack oak bookcase, cradle to gate. The factory's export manager had negotiated fabric specifications and shipping schedules for eleven years. He had no ready answer. The company knew how much electricity it purchased and roughly how much diesel its forklifts burned. It did not know the emissions embedded in the particleboard it bought from a supplier two hundred kilometers away, the urea-formaldehyde resin in its veneer, or the freight from Shenzhen to Rotterdam. The questionnaire asked for a number in kilograms of carbon dioxide equivalent, to two decimal places. The factory could produce a number. Whether that number meant anything is a different question.

This chapter is about the gap between the demand for a number and the capacity to produce one that is defensible. Green governance, as the previous chapter argued, is a capability rather than a certificate. But capabilities are invisible until they are measured. Carbon accounting and monitoring, reporting, and verification (MRV) systems are the instruments through which the invisible becomes visible. They are also the instruments through which firms learn. Measurement is not a neutral recording of what already exists. It is an organizational act that creates new knowledge, new categories of cost, and new possibilities for coordination. The same Foshan factory that cannot answer the buyer's questionnaire today may, after three years of building an accounting system, discover that its largest emissions source is not the diesel it already knew about but the resin it never considered. That discovery is a governance event. It changes what the firm can see, and therefore what it can manage.

3.1 Carbon Accounting Standards: Scope 1, 2, and 3 Emissions

3.1.1 The Genealogy of a Classification

The three-scope architecture that now organizes virtually all corporate carbon accounting emerged from a specific institutional moment. In the late 1990s, as the Kyoto Protocol negotiations advanced, a group of corporations and environmental organizations began meeting under the auspices of the World Resources Institute and the World Business Council for Sustainable Development. Their problem was practical. Companies were beginning to report emissions, but they reported different things in different ways. A cement company might count the carbon dioxide released when limestone is calcined in its kilns. An electricity utility might count the emissions from its smokestacks. A retail chain might count nothing, because it owned no smokestacks at all. Without a common frame, the numbers were incommensurable.

Published in 2001 and revised in 2004, the Greenhouse Gas Protocol solved this problem by distinguishing emissions according to where they occur in the value chain relative to the reporting firm's ownership boundary. Scope 1 covers direct emissions from sources the firm owns or controls: fuel combustion in boilers and furnaces, process emissions from chemical reactions, fugitive emissions from refrigerants and methane leaks. Scope 2 covers indirect emissions from purchased electricity, steam, heating, and cooling, which occur at a power plant but are attributable to the firm that consumed the power. Scope 3 covers all other indirect emissions in the value chain, upstream and downstream: purchased goods and services, business travel, employee commuting, transportation and distribution, processing of sold products, use of sold products, end-of-life treatment.

The classification is elegant in its simplicity and treacherous in its application. Scope 1 and 2 have clear ownership boundaries. Scope 3 does not. The boundary of Scope 3 is, in principle, the entire economy. A furniture factory's Scope 3 emissions include the emissions from growing the trees, harvesting them, processing them into board, manufacturing the hardware, generating the electricity used by its suppliers, transporting everything, and eventually disposing of the bookcase when the consumer tires of it. No firm can measure all of this directly. Scope 3 accounting is therefore an exercise in estimation, and estimation is an exercise in judgment about where to stop.

3.1.2 Scope 1: The Illusion of Direct Control

Scope 1 emissions have the appearance of being the easiest to measure. The firm owns the boiler; the boiler burns gas; the gas bill records the volume. A straightforward calculation converts cubic meters of natural gas into tons of carbon dioxide. The emission factors are published by the Intergovernmental Panel on Climate Change and updated periodically. What could be simpler?

The simplicity is real but partial. Consider a Chinese ceramics manufacturer in Foshan's Shiwan district, where kilns have fired porcelain for five centuries. The kilns burn natural gas, and the gas meter gives a clean number. But the kilns also release carbon dioxide from the decomposition of carbonates in the clay and glazes, process emissions that no fuel meter captures. In 2019, a mid-sized tile manufacturer in the region discovered, through a mandatory emissions audit, that process emissions accounted for roughly 18 percent of its Scope 1 total. The firm had been reporting only combustion. It had not been lying. It had been measuring what its meters measured.

This pattern recurs across industries where chemical reactions, not just fuel combustion, release greenhouse gases. Cement production is the canonical case: roughly 60 percent of the sector's emissions come from the calcination of limestone, not from the fuel used to heat the kiln. A cement plant that measures only its coal consumption will understate its emissions by a factor of two or more. The same is true, in different proportions, for steel (where coke used as a reducing agent, not just as fuel, releases carbon), aluminum (where anode consumption releases carbon dioxide), and glass (where carbonate batch materials decompose).

Scope 1 accounting requires a theory of the production process, not just a meter. The firm must know which chemical reactions occur, in what quantities, and with what stoichiometry. This knowledge is often held by process engineers who have never been asked to think of their production line as a carbon source. Building Scope 1 accounting capability means translating engineering knowledge into environmental accounting categories. The translation is not automatic. In our fieldwork, we encountered a specialty chemicals plant that had reported zero process emissions for three years because the engineer responsible for the emissions inventory had classified the relevant reaction as "part of the product, not a waste." Technically, the carbon dioxide was captured and sold to a beverage company. But the capture rate was 87 percent, and the remaining 13 percent vented to the atmosphere. The inventory said zero. The atmosphere said otherwise.

3.1.3 Scope 2: The Location versus Market Problem

Scope 2 appears even simpler than Scope 1: the firm buys electricity, the utility reports its emission factor, the firm multiplies kilowatt-hours by kilograms of carbon dioxide per kilowatt-hour. The complication arises from the fact that electricity grids are not homogeneous. The carbon intensity of a kilowatt-hour depends on when and where it was generated. A factory in Sichuan, dominated by hydropower, consumes electricity with a carbon intensity roughly one-fifth that of a factory in Shanxi, where coal dominates. A factory that operates only during daylight hours may consume a higher share of solar power than one that runs three shifts.

The Greenhouse Gas Protocol offers two methods for Scope 2 accounting. The location-based method uses the average emission factor of the grid in the region where the electricity is consumed. The market-based method uses the emission factor of the electricity the firm has contractually purchased, including renewable energy certificates and power purchase agreements. The two methods can produce wildly different numbers. A factory in Guangdong that purchases renewable energy certificates can report near-zero market-based Scope 2 emissions while its location-based emissions remain substantial. Neither number is wrong. They answer different questions. The location-based number answers: what is the carbon intensity of the grid this factory draws from? The market-based number answers: what is the carbon intensity of the electricity this factory has paid for?

The choice between methods is strategic rather than technical. A firm seeking to demonstrate progress to investors will prefer the market-based method, which allows it to claim credit for renewable procurement. A firm seeking to understand its physical exposure to carbon pricing will prefer the location-based method, which reflects the actual grid it depends on. A regulator designing a carbon market must choose which method to use for compliance, and that choice will determine whether renewable procurement counts as an offset against obligations. In China's national emissions trading system, the rules have evolved toward a hybrid approach: enterprises report both, but compliance obligations are calculated using grid-average factors adjusted for certain recognized green power purchases. The adjustment rules are technical. Their distributional consequences are not. They determine who pays for the transition.

3.1.4 Scope 3: The Boundary That Cannot Hold

Scope 3 is where carbon accounting becomes genuinely difficult, and where the governance stakes are highest. For most manufacturing firms, Scope 3 emissions dwarf Scope 1 and 2 combined. For a furniture company, the emissions embedded in purchased materials (board, metal hardware, finishes, packaging) typically account for 70 to 90 percent of the total carbon footprint. For an electronics assembler, the share is similar. For an automobile manufacturer, the use-phase emissions of sold vehicles can be five to ten times the manufacturing emissions. The numbers are large, and they are mostly estimates.

The Greenhouse Gas Protocol's Scope 3 standard, published in 2011, identifies fifteen categories of Scope 3 emissions and provides guidance on calculation methods. The methods range from direct measurement (rare) to supplier-specific data (increasingly common) to industry-average emission factors (the default). The choice of method matters enormously. A furniture company that uses an industry-average emission factor for particleboard will get a different number than one that obtains supplier-specific data from its board mill. The difference can be a factor of two or more, depending on the mill's energy source, the efficiency of its dryers, and the share of recycled versus virgin fiber.

The deeper problem is that Scope 3 accounting requires the firm to know things it has no authority to know. It requires knowing how its suppliers produce, how its customers use, and how its products are disposed of. The firm can ask. It can require suppliers to report. It can commission life cycle assessment studies. But it cannot verify any of this with the confidence it can verify its own gas meter. Scope 3 is, by construction, an exercise in boundary-setting under uncertainty.

Consider the furniture case in more detail. A Chinese furniture exporter selling to IKEA or a European brand will be asked to report the carbon footprint of each product. The calculation follows the ISO 14067 standard for product carbon footprints, which is built on life cycle assessment methodology. The system boundary is typically cradle-to-gate: from raw material extraction through production, excluding distribution, use, and end-of-life. Within that boundary, the largest contributor is usually the board, whether particleboard, medium-density fiberboard, or plywood. The board's carbon footprint depends on the resin used (urea-formaldehyde has a higher footprint than phenol-formaldehyde or MDI), the energy source of the board mill (natural gas, biomass, coal), and the transport distance from mill to factory.

A furniture factory in Foshan that sources board from a mill in Guangxi, 600 kilometers away, may find that transport accounts for 8 percent of the product footprint. If it switches to a mill in Guangdong, 150 kilometers away, transport falls to 2 percent, but the closer mill may use coal-fired dryers while the distant mill uses natural gas. The net effect on the carbon footprint depends on the relative magnitudes, which the factory cannot know without data from both mills. The factory's Scope 3 boundary is, in practice, a network of data dependencies it does not control.

This is the central insight that Scope 3 accounting forces on firms: the boundary of the firm's carbon responsibility extends far beyond the boundary of its legal control. A firm can choose its suppliers, but it cannot choose their energy systems. It can specify materials, but it cannot specify the processes that made them. Scope 3 accounting makes visible a set of dependencies that were always there but were never measured. Whether making them visible leads to better governance depends on whether the firm has the leverage and the incentive to act on what it sees.

3.1.5 The Politics of Scope Allocation

The three-scope architecture is often presented as a technical convention. It is also a political settlement. By assigning emissions to scopes, the protocol distributes responsibility among actors in the value chain. Scope 1 emissions are the responsibility of the firm that owns the emitting asset. Scope 2 emissions are the responsibility of the firm that consumes the electricity, though the power plant is the physical source. Scope 3 emissions are the responsibility of the firm that purchases the goods, though the supplier is the physical source.

This allocation has consequences. A coal-fired power plant reports its emissions as Scope 1. The factories that buy its electricity report the same emissions as Scope 2. The emissions are counted twice, once by the plant and once by the factory. The Greenhouse Gas Protocol acknowledges this double counting and accepts it as a feature, not a bug. The purpose of corporate accounting is not to produce a global total (that is the job of national inventories) but to assign responsibility to actors who can act. The power plant can act by changing its fuel. The factory can act by changing its supplier or by investing in efficiency. Double counting allows both to be held accountable.

Double counting also creates opportunities for arbitrage. A firm can reduce its reported Scope 2 emissions by purchasing renewable energy certificates without any change in the physical electricity it consumes. The emissions have not disappeared; they have been reallocated to someone else who did not purchase the certificates. Whether this reallocation represents a real reduction depends on whether the certificates finance additional renewable generation. The additionality question, familiar from the carbon offset literature, reappears at the level of corporate accounting. The Greenhouse Gas Protocol's market-based method counts certificate purchases as reductions. Critics argue that this allows firms to claim progress without changing their operations. The debate is unresolved, and it is not primarily a technical debate. It is a debate about what counts as a real change.

For Chinese manufacturers, the politics of scope allocation has a particular edge. Chinese firms are embedded in global value chains as suppliers. Their Scope 1 and 2 emissions are, from the perspective of their foreign buyers, Scope 3 emissions. When a European retailer reports its Scope 3 emissions, it is reporting the emissions of its Chinese suppliers. The retailer has an incentive to reduce its reported Scope 3 emissions. It can do so by pressuring suppliers to reduce their emissions, by switching to suppliers with lower emissions, or by reclassifying certain activities as outside its boundary. The supplier bears the cost of the first option, the risk of the second, and the exclusion of the third. The scope architecture, which appears to be a neutral accounting convention, structures a power relationship.

3.1.6 What Scopes Reveal and Conceal

The three-scope architecture reveals certain things and conceals others. It reveals the direct emissions of the reporting firm, which is useful for understanding its exposure to carbon pricing. It reveals the emissions of the electricity it consumes, which is useful for understanding its exposure to grid decarbonization. It reveals, to the extent that estimates are reliable, the emissions embedded in its value chain, which is useful for understanding its exposure to supply chain pressures.

It conceals the emissions of other firms that are not the reporting firm's suppliers but are part of the same industrial ecosystem. A furniture factory's Scope 3 includes its board supplier but not the board supplier's competitors, who may be more or less efficient. It conceals the emissions associated with the infrastructure the firm uses (the roads, ports, and transmission lines that make its operations possible) because these are typically treated as background. It conceals the emissions avoided by the firm's products, such as a building insulation material that reduces heating energy use, because avoided emissions are not part of the accounting framework. And it conceals the distribution of emissions within the value chain, because the reporting firm aggregates its suppliers into a single number.

These concealments are not accidents. They reflect the purpose for which the protocol was designed: to help individual firms measure and manage their own emissions. The protocol was not designed to support industrial policy, regional planning, or value chain governance. When it is used for those purposes, as it increasingly is, its blind spots become consequential. A regional government that wants to reduce the carbon intensity of its furniture industry cannot rely on firm-level Scope 3 reports, because those reports aggregate suppliers in ways that obscure the regional picture. A value chain organizer that wants to identify the highest-leverage intervention points cannot rely on Scope 1 and 2 alone, because the largest emissions are often in Scope 3. The architecture is a starting point, not a complete map.

3.2 MRV Systems: Design, Implementation, and Challenges

3.2.1 The Institutional Machinery of Credible Numbers

A carbon number is only as good as the system that produces it. Monitoring, reporting, and verification (MRV) is the institutional machinery that transforms a firm's self-reported estimate into a number that regulators, investors, and buyers can trust. The three components are distinct. Monitoring is the ongoing collection of data: meter readings, fuel purchases, production volumes, emission factors. Reporting is the compilation of that data into a standardized format: an emissions inventory, a sustainability report, a product carbon footprint declaration. Verification is the independent check: a third-party auditor reviews the data, the methodology, and the controls to confirm that the reported number is accurate and complete.

The logic of MRV is borrowed from financial auditing. Just as investors rely on audited financial statements, carbon market participants rely on verified emissions reports. The analogy is instructive but imperfect. Financial auditing has centuries of precedent, standardized accounting rules, and a professional cadre of auditors. Carbon MRV is younger, its rules are still evolving, and its auditors are fewer. The International Accreditation Forum estimates that fewer than 500 bodies worldwide are accredited to verify greenhouse gas inventories, and their distribution is highly uneven. Europe and North America have the densest coverage. China has a growing cadre, but the capacity is concentrated in a few provinces and a few large firms.

3.2.2 Monitoring: The Data Collection Problem

Monitoring begins with data. The data comes from meters, invoices, production records, and supplier declarations. Each source has its own reliability characteristics. A utility bill is generally reliable: it records the kilowatt-hours delivered and billed. A fuel purchase invoice is reliable for quantity but may not specify the carbon content of the fuel, which varies by source. A production record is reliable for output but may not capture the process conditions that determine emissions intensity. A supplier declaration is the least reliable: it depends on the supplier's own monitoring system, which may be rudimentary.

The monitoring challenge is particularly acute for small and medium-sized enterprises. A large steel mill has continuous emissions monitoring systems on its stacks, calibrated to regulatory standards. A small furniture factory has a utility meter and a fuel receipt. The difference in data quality is not a difference in virtue; it is a difference in resources. The steel mill can afford the monitoring equipment because its compliance obligations justify the investment. The furniture factory cannot, because its obligations are lighter and its margins thinner.

China's national carbon market has addressed this asymmetry by focusing initially on large emitters. When the market launched in 2021, it covered 2,225 enterprises in the power sector, each emitting more than 26,000 tons of carbon dioxide per year. These enterprises are required to install monitoring equipment, submit annual emissions reports, and have those reports verified by third-party agencies. The threshold excludes most manufacturing firms, including most furniture factories. For those firms, monitoring is voluntary, driven by buyer requirements rather than regulatory mandate. The result is a two-tier system: a rigorous MRV regime for large emitters, and a patchwork of voluntary initiatives for everyone else.

3.2.3 Reporting: Standardization and Its Discontents

Reporting transforms monitoring data into a standardized disclosure. The dominant standards are the Greenhouse Gas Protocol's Corporate Standard, the ISO 14064 series, and, for product-level reporting, ISO 14067 and the Product Environmental Footprint guidelines. These standards specify what to report, how to calculate it, and how to present it. They do not specify the software, the organizational structure, or the internal controls that support the calculation. Those are left to the firm.

The standardization of reporting has advanced rapidly. The Task Force on Climate-related Financial Disclosures, established by the Financial Stability Board in 2015, produced a framework that has been widely adopted by listed companies. The International Sustainability Standards Board, created in 2021, is consolidating various voluntary frameworks into a global baseline. China's securities regulator has begun requiring listed companies to disclose environmental information, though the requirements remain less detailed than those in the European Union.

Standardization solves the comparability problem at the level of format. It does not solve it at the level of substance. Two firms can both report in accordance with the Greenhouse Gas Protocol and produce numbers that are not comparable, because they have made different choices about boundary-setting, emission factors, and estimation methods. The protocol allows firms to choose their organizational boundary (equity share, financial control, or operational control) and their operational boundary (which scopes to include). A firm that reports on an equity-share basis will report different numbers than a firm that reports on an operational-control basis, even if their physical operations are identical. The protocol requires firms to disclose their boundary choices, but disclosure does not make the numbers comparable. It merely allows a careful reader to understand why they are not.

3.2.4 Verification: The Third-Party Assurance Industry

Verification is the component of MRV that has attracted the most scrutiny, and for good reason. The credibility of the entire system depends on the independence and competence of verifiers. If verifiers are captured by the firms they audit, or if they lack the technical capacity to detect errors, the reported numbers are unreliable regardless of how rigorous the monitoring and reporting standards are.

The verification industry has grown rapidly. The Clean Development Mechanism, established under the Kyoto Protocol, accredited designated operational entities to validate projects and verify emission reductions. The voluntary carbon market has spawned a parallel set of standards and verifiers, including Verra, Gold Standard, and the Climate Action Reserve. The European Union's emissions trading system requires verification by accredited bodies under the European Cooperation for Accreditation. China's national carbon market requires verification by agencies registered with the Ministry of Ecology and Environment.

The quality of verification varies. A 2019 study by the European Court of Auditors found that the EU ETS verification system had weaknesses, including insufficient checks on whether verifiers were independent and whether they had the technical competence to assess complex industrial processes. A 2020 investigation by Carbon Market Watch found that some Chinese verifiers had limited experience with the industries they were auditing. These findings do not mean that verification is worthless. They mean that verification is an institutional capability that takes time and resources to build, and that its quality cannot be assumed.

One structural feature of the verification market makes this harder. The verifier's client is the firm being verified: the firm pays the verifier, and the firm can choose among competing verifiers. This creates an incentive for leniency, just as it does in financial auditing. The standard solution is accreditation and oversight, in which an independent body accredits verifiers, monitors their performance, and can revoke their accreditation. The effectiveness of this solution depends on the resources and independence of the accrediting body. In jurisdictions where regulatory capacity is thin, the solution is only as strong as the regulator.

3.2.5 MRV in China: The Infrastructure Build-Out

China's MRV infrastructure has expanded rapidly since the national carbon market was announced in 2017 and launched in 2021. The Ministry of Ecology and Environment has issued a series of technical guidelines covering accounting and reporting for key industries: power generation, steel, cement, aluminum, and others. These guidelines specify the monitoring methods, emission factors, and reporting formats that enterprises must use. They are, in effect, a national carbon accounting standard.

The implementation of these standards has been uneven. A 2022 review by the Ministry found that some enterprises had submitted reports with calculation errors, missing data, and inconsistent boundary definitions. The Ministry responded by tightening the verification requirements and increasing the frequency of spot checks. The review also found that the capacity of local environmental protection bureaus to oversee MRV varied widely, with wealthier coastal provinces better equipped than inland provinces.

The MRV system is also being extended beyond the national carbon market. The Ministry has piloted a voluntary greenhouse gas reporting program for enterprises outside the mandatory market. Some provincial governments have established their own reporting requirements. The China Emissions Exchange in Guangzhou and the Shanghai Environment and Energy Exchange have developed verification protocols for voluntary offsets. The infrastructure is being built piece by piece, driven by the twin pressures of domestic climate policy and international trade requirements.

3.2.6 The MRV-Capability Nexus

The relationship between MRV and green governance capability is recursive. A firm that participates in MRV learns things it did not know. It learns where its emissions come from, which processes are most carbon-intensive, and which suppliers are most exposed. This knowledge is a precondition for action. But the knowledge is also shaped by the MRV system itself. The categories the system uses, including scopes, emission factors, and activity data, determine what the firm can see. A system that requires reporting only Scope 1 and 2 will not reveal the emissions embedded in the value chain. A system that uses industry-average emission factors will not reveal the differences between suppliers. The MRV system is not a window onto reality; it is a frame that selects certain aspects of reality for attention.

This framing effect has implications for industrial upgrading. If the MRV system rewards firms for reducing Scope 1 emissions, firms will invest in combustion efficiency and fuel switching. If it rewards them for reducing Scope 3 emissions, they will invest in supply chain management and material substitution. If it rewards them for reporting accuracy, they will invest in monitoring equipment and data systems. The direction of investment follows the structure of the measurement system. Getting the measurement system right is therefore not a technical afterthought. It is a central element of green industrial policy.

3.3 Boundaries of Evidence: Data Gaps, Uncertainty, and Comparability

3.3.1 The Data Gap Problem

The most basic limitation of green governance measurement is that the data often do not exist. A firm cannot report what it has not measured. A regulator cannot verify what has not been reported. A researcher cannot analyze what has not been collected. The data gaps are not random. They are systematically distributed across firm size, industry, and geography.

Small firms are less likely to have emissions data than large firms. The reasons are straightforward: small firms lack the staff, the equipment, and the regulatory incentive to measure. A 2022 survey by the China Association of Small and Medium Enterprises found that fewer than 15 percent of small manufacturing firms had conducted a carbon footprint assessment, compared with 68 percent of large firms. The gap is even wider in inland provinces than in coastal ones.

Informal firms are almost entirely absent from the data. China's manufacturing sector includes a large number of small workshops and household enterprises that operate outside the formal reporting system. These firms may be significant emitters in aggregate, but they are invisible in firm-level data. The furniture industry is a case in point: alongside the large export-oriented factories are thousands of small workshops that supply components, perform finishing, and assemble products. Their emissions are real, but they are not counted.

The data gap is also a value chain gap. A firm may have good data on its own operations but poor data on its suppliers. The suppliers may be small, informal, or simply unwilling to share. The result is that Scope 3 estimates rely heavily on industry averages, which may be far from the actual emissions of any particular supplier. The furniture factory that uses an industry-average emission factor for particleboard may be off by 50 percent or more, depending on whether its actual supplier uses biomass or coal for drying.

3.3.2 Uncertainty: The Error Bars Nobody Reports

Every carbon number has an error bar. The error bar is rarely reported. A firm that reports 1,000 tons of Scope 1 emissions may be accurate to within 5 percent, or it may be accurate to within 50 percent, depending on the quality of its meters and the completeness of its process knowledge. A Scope 3 estimate may be accurate to within a factor of two. The convention of reporting a single number without an uncertainty range creates an illusion of precision that the underlying data do not support.

The sources of uncertainty are multiple. Measurement uncertainty arises from the accuracy of meters and the variability of emission factors. Estimation uncertainty arises from the use of proxies and averages. Boundary uncertainty arises from decisions about what to include and what to exclude. Model uncertainty arises from assumptions about future conditions, such as the grid emission factor over the lifetime of a product.

Quantifying uncertainty is possible but rarely done. The IPCC guidelines provide methods for estimating uncertainty ranges, and the Greenhouse Gas Protocol recommends that firms disclose their uncertainty. In practice, few firms do. The reasons are partly technical (uncertainty analysis requires expertise) and partly strategic (disclosing uncertainty weakens the rhetorical force of a number). A firm that reports "1,000 tons, plus or minus 300" is less persuasive than a firm that reports "1,000 tons." The incentive is to suppress the error bar.

The suppression of uncertainty has consequences for governance. If a regulator sets a reduction target based on a reported number that is uncertain, the target may be met or missed for reasons unrelated to actual emissions. If an investor evaluates a firm's climate performance based on reported numbers that are not comparable, the evaluation may reward reporting sophistication rather than emission reduction. If a buyer selects suppliers based on carbon footprint declarations that use different methodologies, the selection may favor suppliers with lenient accountants rather than low emissions.

3.3.3 Comparability: The Problem of Apples and Oranges

Comparability is the ability to compare carbon numbers across firms, products, or time periods. It is the foundation of any governance system that uses measurement to allocate rewards or penalties. A carbon market requires comparable emissions reports to set allowances. A green procurement program requires comparable product footprints to select suppliers. A climate disclosure regime requires comparable corporate reports to inform investors.

Comparability is harder to achieve than standardization. Two firms can follow the same standard and still produce incomparable numbers, because the standard allows choices. The Greenhouse Gas Protocol allows firms to choose their organizational boundary, their operational boundary, their emission factors, and their estimation methods. The ISO 14067 standard for product carbon footprints allows choices about system boundaries, allocation methods, and data quality. These choices are not trivial. They can change the reported number by a factor of two or more.

The comparability problem is particularly acute for product carbon footprints, which are increasingly used in green procurement. A European retailer that asks its Chinese suppliers to report product footprints will receive numbers calculated according to different standards, different boundaries, and different data sources. The retailer may try to harmonize the numbers by specifying a standard and a boundary. But even with harmonization, the numbers may not be comparable, because the underlying data quality varies. A supplier with a sophisticated life cycle assessment team will produce a more accurate number than a supplier using a spreadsheet and industry averages. The more accurate number may be higher or lower than the less accurate one. The retailer cannot tell which is which without auditing both.

3.3.4 The Evidence Hierarchy

Not all evidence is equal. A hierarchy of evidence for green governance measurement would place direct measurement at the top, followed by supplier-specific data, followed by industry averages, followed by proxy estimates. The hierarchy is not absolute: a well-constructed industry average may be more reliable than a poorly executed direct measurement. But in general, the further the evidence is from the actual emission source, the more uncertain it is.

The evidence hierarchy has implications for the argument of this book. If green governance capability is to be an endogenous driver of upgrading, firms must be able to generate evidence about their own performance and the performance of their suppliers. Firms that rely on industry averages are not governing; they are guessing. Firms that invest in direct measurement and supplier-specific data are building the evidence base for governance. The investment in measurement is itself a form of capability building.

This is why the MRV system matters beyond compliance. A firm that participates in a rigorous MRV system learns to generate evidence. It develops the internal systems, the staff expertise, and the data infrastructure that allow it to govern its emissions. The learning is not automatic, since a firm can comply with MRV requirements without changing its operations, but the capability is a precondition for more ambitious action. You cannot manage what you cannot measure. You cannot improve what you cannot verify.

3.3.5 The Limits of Measurement

Measurement has limits that are not technical. Some things that matter for green governance are not easily measured. The environmental impact of a product depends not only on its carbon footprint but on its water use, its land use, its toxicity, and its contribution to biodiversity loss. These impacts are harder to quantify than carbon, and they are often omitted from governance systems that focus on carbon alone. A furniture factory that switches from solid wood to particleboard may reduce its carbon footprint (because particleboard uses waste wood and requires less energy to produce) while increasing its use of adhesives that contain volatile organic compounds. The carbon number improves; the overall environmental performance may not.

Measurement also has distributional limits. The firms that can afford to measure are the firms that can afford to comply. Small firms, informal firms, and firms in low-income regions are less able to participate in measurement systems. If green governance rewards firms that measure well, it may also reward firms that are already advantaged. The measurement system can become a barrier to entry, not a level playing field.

These limits do not invalidate measurement. They qualify it. Measurement is necessary but not sufficient for green governance. It provides the information that governance requires, but it does not provide the authority, the incentive, or the coordination that governance also requires. The chapters that follow examine how firms, industries, and regions use measurement, and sometimes fail to use it, as they navigate the transition to green competitiveness.

3.3.6 The Foshan Factory Revisited

Return to the furniture factory in Foshan. Three years after receiving the buyer's questionnaire, the factory has built a carbon accounting system. It has installed submeters on its main production lines. It has obtained supplier-specific emission factors from its two largest board suppliers. It has conducted a life cycle assessment of its best-selling bookcase, following ISO 14067. The assessment took eight months and cost roughly 120,000 yuan, paid partly by the factory and partly by a grant from the provincial government.

The result was surprising. The factory's Scope 1 and 2 emissions were lower than expected: the biggest source was not the drying kilns but the diesel forklifts, which idled for hours between shifts. The Scope 3 emissions were higher than expected: the particleboard supplier used coal-fired dryers, and the resin supplier used a process that released significant methane. The total footprint of the bookcase was 40 percent higher than the industry average the factory had been using.

The factory now faced a choice. It could switch board suppliers to a mill that used natural gas dryers, which would reduce the footprint by 25 percent but increase the board cost by 8 percent. It could invest in electric forklifts, which would reduce Scope 1 emissions by 15 percent but require a charging infrastructure and a change in shift patterns. It could reformulate the resin, which would require working with the chemical supplier to develop a lower-emission formulation. Each option required investment, and each investment had a payback period that depended on whether the buyer would pay a premium for a lower-carbon product.

The factory's manager told us that the measurement exercise had changed his understanding of his own business. "We thought we were a wood products company," he said. "Now we know we are a carbon company that happens to make furniture." The statement is an exaggeration, but it captures something real. The act of measurement had reframed the firm's identity. It had made visible a set of relationships, with suppliers, with energy systems, with customers, that had always been there but had never been the object of management attention. Whether that reframing translates into a competitive advantage depends on what the firm does next. But the reframing is a precondition. The firm cannot govern what it cannot see.

3.4 Conclusion: Measurement as Capability

Measurement is not a prelude to green governance; it is a form of green governance. The systems that produce carbon numbers, including the accounting standards, the MRV infrastructure, and the verification industry, are not neutral tools that record an independent reality. They are institutions that create the categories through which firms understand their environmental performance. They determine what is visible and what is invisible, what is counted and what is excluded, what is rewarded and what is ignored.

Three implications follow for the framework developed in this book.

Green governance capability includes measurement capability. A firm that cannot produce a credible carbon number cannot participate in the governance systems that increasingly structure global value chains. It cannot respond to buyer questionnaires, comply with carbon market obligations, or access green finance. The inability to measure is a form of exclusion.

The measurement systems themselves are sites of capability building. A firm that builds a carbon accounting system learns about its production processes, its supply chain, and its energy use. The learning is not incidental; it is the mechanism through which measurement translates into capability. The Foshan factory did not just produce a number. It produced a new understanding of its business.

The boundaries of evidence are also the boundaries of governance. The data gaps, uncertainties, and comparability problems described in this chapter are not temporary technical difficulties that will be solved by better software. They are structural features of a measurement system that was designed for firm-level accounting, not for value chain governance. The system can be improved, but it cannot be made perfect. Governance must proceed under conditions of imperfect evidence.

The next chapter turns to the policy architecture that shapes the incentive landscape for Chinese manufacturers. The dual carbon goals, peak emissions before 2030 and carbon neutrality before 2060, are the most ambitious climate commitment any developing country has made. They are also a governance challenge of enormous complexity. The measurement systems described in this chapter are the infrastructure through which that challenge will be met, or not. The chapters that follow examine how different industries, labor-intensive, technology-intensive, and capital-intensive, are building the measurement capabilities they need, and what happens when they cannot.

The evidence will not be uniform. Some firms will build sophisticated systems; others will rely on estimates. Some industries will develop collective measurement infrastructure; others will leave it to individual firms. The variation is not a failure of the framework. It is the data the framework must explain. Measurement capability is unevenly distributed because green governance capability is unevenly distributed, and the two are not separable.

There is a temptation, when writing about measurement, to treat it as a preliminary stage, something that must be gotten out of the way before the real work of governance begins. The temptation should be resisted. In the Chinese manufacturing sectors examined in this book, the firms that measure well are, almost without exception, the firms that govern well. The correlation is not perfect; some firms measure diligently and then do nothing with the numbers, while others govern effectively through informal knowledge that never appears in a spreadsheet. But the correlation is strong enough to suggest that measurement is one of the mechanisms through which capability is produced, not merely a symptom of it.

Consider what happens inside a firm when it builds a carbon accounting system for the first time. The process begins with a boundary decision: which facilities, which processes, which gases. That decision forces the firm to articulate what it means by "the firm," a question that legal incorporation answers for tax purposesbut not for environmental purposes. A company with three factories, two joint ventures, and a network of contract manufacturers must decide which of these entities sit inside its carbon boundary. The decision is not obvious, and different members of the management team will have different views. The debate itself is instructive. It reveals that the firm's operational reality is more complex than its organizational chart suggests.

Once the boundary is set, the firm must map its energy flows. This is where the learning becomes concrete. A furniture factory that has always paid its electricity bill without examining it will discover, through the mapping exercise, that its drying kilns account for 60 percent of consumption, its dust extraction system for 20 percent, and everything else for the remaining 20 percent. The discovery is not new information in a strict sense, since the meters have always been there, but it is new knowledge in an organizational sense. Nobody had assembled the data in a way that made the pattern visible. The assembly is the capability.

The mapping exercise also reveals dependencies the firm had not considered. The factory discovers that its largest supplier of particleboard is itself a significant energy consumer, and that the supplier's energy source (coal, gas, or biomass) determines a large share of the factory's own footprint. The factory cannot control the supplier's energy choice, but it can influence it through procurement decisions, technical assistance, or joint investment. Measurement has created a new object of management: the supplier's energy system. Before the measurement, the supplier was a source of board at a price. After the measurement, the supplier is a source of carbon at a price. The relationship has acquired a new dimension.

This is what it means to say that measurement is constitutive of governance. It does not merely record a relationship that already exists. It creates a new relationship by making the old one visible in a new way. The firm that measures its supply chain carbon is not the same firm that did not measure it, even if its physical operations are unchanged. Its managers see a different set of problems, and they have a different set of tools for addressing them.

3.4.1 The Institutional Dimension of Measurement Capability

Measurement capability is not only a firm-level attribute. It is also an institutional attribute, distributed across the public and private organizations that produce the standards, verify the reports, and train the personnel. A firm's ability to measure depends on the availability of emission factors, the existence of accredited verifiers, the presence of trained accountants and engineers, and the credibility of the regulatory framework that requires or encourages measurement. These are collective goods, not private ones. A single firm cannot produce them on its own.

China's measurement infrastructure has developed rapidly, but it remains uneven. The national carbon market has created a cadre of trained verifiers and a set of technical guidelines that firms can use. The Ministry of Ecology and Environment has published emission factors for major fuels and industrial processes. The China National Institute of Standardization has issued standards for carbon footprint calculation. These are genuine achievements, built in less than a decade.

The unevenness is geographic and sectoral. The verifiers are concentrated in Beijing, Shanghai, Guangzhou, and Shenzhen. The emission factors are most detailed for the industries covered by the national carbon market (power, steel, cement, aluminum) and less detailed for industries outside it, including furniture. The training programs are accessed disproportionately by large firms with the resources to send staff to workshops. A small furniture factory in an inland province may have no access to any of this infrastructure except through a consultant, and the consultant may be more expensive than the factory can afford.

The institutional dimension matters for the argument of this book because it shapes the distribution of green governance capability. Firms located in regions with dense measurement infrastructure, and operating in sectors covered by the national carbon market, have a structural advantage in building measurement capability. Firms located elsewhere, or operating in uncovered sectors, must build capability with less support. The difference is not a difference in managerial talent. It is a difference in the institutional environment.

This has implications for industrial policy. If the goal is to build green governance capability across the manufacturing sector, not just in the large emitters covered by the carbon market, then the measurement infrastructure must be extended. Emission factors must be developed for more industries. Verifiers must be trained and accredited in more regions. Standards must be adapted for small firms that cannot afford full life cycle assessment. The extension is not glamorous work. It does not produce ribbon-cutting ceremonies or headline announcements. But it is the foundation on which everything else rests.

3.4.2 Measurement and the Relocation Question

The boundaries of evidence have a direct bearing on the relocation question that runs through this book. If Chinese manufacturers cannot produce credible carbon numbers, they will face pressure from buyers who need those numbers for their own reporting. The pressure can be resolved in three ways. The manufacturer can build measurement capability and report accurate numbers. The buyer can accept estimates based on industry averages, which reduces the credibility of the buyer's own reporting. Or the buyer can switch to a supplier in a jurisdiction where measurement is easier, which may mean relocating production to a country with a more developed MRV infrastructure.

The third option is the one that concerns Chinese policymakers. A furniture factory in Vietnam that can produce a verified carbon footprint may be more attractive to a European buyer than a factory in China that cannot, even if the Chinese factory's actual emissions are lower. The buyer is not choosing between actual emissions; the buyer is choosing between reported emissions. The difference between actual and reported is the measurement gap, and the measurement gap is a competitive disadvantage.

This is not a hypothetical. In interviews conducted for this book, procurement managers at two European retailers said that they had begun to weight carbon reporting capability in supplier selection. One manager described a scoring system in which suppliers received points for verified footprints, for renewable energy procurement, and for participation in industry benchmarking initiatives. The points were not the only criterion, since price, quality, and delivery reliability remained dominant, but they were a tiebreaker. A supplier that could not report lost to a supplier that could, even when the underlying products were comparable.

Measurement capability is becoming a dimension of export competitiveness. In the first two decades of China's export boom, competitiveness was defined by cost, quality, and speed. Carbon reporting was not part of the equation. Now it is, at least for exports to markets with stringent climate disclosure requirements. The shift is gradual, and it is uneven across product categories and destination markets. But the direction is clear.

The relocation question, then, is not only about where production physically occurs. It is about where the evidence of production is produced. A factory that relocates its assembly operations to another country but retains its measurement systems in China may be able to satisfy buyers without moving the emissions-intensive processes. A factory that cannot measure will find it harder to retain any part of the value chain that requires carbon disclosure. The geography of measurement is becoming a factor in the geography of production.

3.4.3 The Boundaries of This Chapter's Evidence

It would be inconsistent to write a chapter about the boundaries of evidence without acknowledging the boundaries of the evidence presented here. The claims in this chapter rest on a combination of published standards, secondary literature, and fieldwork conducted in Guangdong, Zhejiang, and Jiangsu provinces between 2019 and 2023. The fieldwork included interviews with factory managers, verification agency staff, provincial environmental officials, and procurement managers at foreign retailers. The sample is not random. It overrepresents firms that are already engaged with carbon measurement, because those are the firms that agreed to be interviewed. It underrepresents firms that have no measurement systems at all, because those firms were harder to identify and less willing to discuss their gaps.

The quantitative data cited in this chapter come from published sources: the Greenhouse Gas Protocol, the IPCC guidelines, the Ministry of Ecology and Environment's technical standards, and industry association reports. These sources are reliable for the standards they describe, but they are not independent evaluations of how those standards are implemented in practice. The implementation evidence comes from the fieldwork and from published audits, which are themselves selective.

These limitations do not invalidate the chapter's argument, but they qualify it. The argument is that measurement capability is constitutive of green governance capability, and that the distribution of measurement capability is uneven in ways that matter for industrial upgrading. The evidence supports this argument. It does not support precise claims about the magnitude of the effect, the speed of its operation, or its variation across subsectors. Those claims would require a larger and more representative dataset than currently exists. The absence of such a dataset is itself a boundary of evidence, one that future research should address.

3.4.4 What the Numbers Can and Cannot Tell Us

The chapters that follow use carbon numbers extensively. They compare the carbon intensity of different industries, the footprints of different products, and the trajectories of different firms. The comparisons are informative, but they must be read with the caveats developed in this chapter in mind. A comparison of carbon intensity across industries is a comparison of reported numbers, not of actual emissions. The reported numbers are shaped by the accounting choices, the data quality, and the verification rigor of the firms that produced them. When a labor-intensive industry appears to have lower carbon intensity than a capital-intensive one, the difference may reflect the industries' actual energy use, or it may reflect the fact that the labor-intensive firms are less likely to have measured their Scope 3 emissions. The two explanations are not mutually exclusive, but they have different implications for policy.

The same caution applies to comparisons across firms within an industry. A firm that reports a lower carbon footprint than its competitor may be more efficient, or it may have chosen a more favorable boundary, or it may have estimated its Scope 3 emissions using a lower emission factor. The reported number alone cannot distinguish between these possibilities. Only the underlying data and methodology can. This is why the MRV system matters: it is the mechanism that allows a reader to distinguish between real differences and accounting differences. Where the MRV system is weak, the reader cannot distinguish, and the comparison is unreliable.

The chapters that follow will flag these limitations where they are material. They will not resolve them, because they cannot be resolved with the available evidence. But they will be explicit about what the numbers can and cannot tell us. That explicitness is itself a contribution. Much of the literature on green competitiveness treats carbon numbers as if they were direct measurements of physical reality. They are not. They are the output of a measurement system, and the system's properties, including its boundaries, its uncertainties, and its comparability, shape what the numbers mean.

3.4.5 Measurement as a Contested Terrain

Measurement is contested. The standards, the methods, and the boundaries are not settled by technical consensus alone. They are shaped by the interests of the actors who produce and use them. Firms prefer accounting rules that minimize their reported emissions. Regulators prefer rules that are verifiable and enforceable. Environmental groups prefer rules that capture the full value chain impact. Investors prefer rules that are comparable across firms. These preferences conflict, and the resolution of the conflict is a political process, not a technical one.

The politics of measurement is visible in the debates over Scope 3. Firms argue that Scope 3 accounting is too uncertain to be reliable, and that including it in mandatory disclosure would impose excessive costs. Environmental groups argue that excluding Scope 3 would let firms off the hook for the majority of their emissions. Regulators are caught between the two, and their positions vary across jurisdictions. The European Union has moved toward mandatory Scope 3 disclosure for large companies under the Corporate Sustainability Reporting Directive. The United States Securities and Exchange Commission proposed a climate disclosure rule that would have required Scope 3 reporting for large firms, but the rule was stayed amid legal challenges. China's approach has been more gradual, with voluntary Scope 3 reporting encouraged but not mandated for most firms.

These differences are not merely technical. They determine which firms bear the cost of measurement, which emissions are visible to governance, and which value chain relationships are subject to scrutiny. A regime that mandates Scope 3 reporting shifts the cost of measurement onto the firms at the downstream end of the value chain, who must collect data from their suppliers. A regime that does not mandate Scope 3 reporting leaves the downstream firms free to report only their direct emissions, which may be a small fraction of their total impact. The choice of regime is a choice about who governs whom.

For Chinese manufacturers, the politics of measurement is experienced as a set of external requirements that arrive through global value chains. The requirements originate in Brussels, Washington, and the headquarters of multinational retailers. They are transmitted to Chinese suppliers through contracts, questionnaires, and audit protocols. The suppliers have little influence over the rules, but they bear the cost of compliance. This asymmetry is a feature of the value chain governance system, and it is unlikely to change in the near term. The best that Chinese manufacturers can do is to build measurement capability that is robust to changes in the rules, capability that can be adapted to new standards as they emerge.

That is the practical conclusion of this chapter. Measurement capability is not a fixed asset. It is a set of organizational routines, data systems, and human skills that can be turned to different purposes as the regulatory environment evolves. A firm that has built a carbon accounting system can adapt it to new standards more easily than a firm that is starting from scratch. A firm that has trained its staff in life cycle assessment can apply those skills to water footprints, circularity metrics, or whatever the next regulatory wave demands. The investment in measurement is an investment in adaptability.

The chapters that follow examine how different industries have made this investment, and what has happened when they have not. The labor-intensive industries, examined in Chapter 5, face the sharpest trade-off between the cost of measurement and the margin available to absorb it. The technology-intensive industries, examined in Chapter 6, have the most to gain from measurement, because their products are increasingly evaluated on environmental performance. The capital-intensive industries, examined in Chapter 7, have the largest emissions and the most developed measurement systems, but also the greatest exposure to the risk of stranded assets. The furniture case, examined in Chapter 9, brings the themes together in a single industry where the measurement gap is wide and the competitive stakes are high.

Before turning to those industries, the next chapter examines the policy architecture that shapes the incentive landscape. The dual carbon goals are the context within which measurement capability is being built. They determine the regulatory requirements, the carbon prices, and the industrial policies that firms must navigate. The measurement systems described here are the infrastructure through which the dual carbon goals will be implemented, or not. The two chapters are complementary: one describes the tools, the other describes the objectives. Together, they set the stage for the industry-level analysis that follows.