Economic history studies how people produced, exchanged, financed and distributed resources in the past, and how those arrangements changed. Its gold standard is an explanation that connects a historical question to evidence, a plausible economic mechanism and an honest account of uncertainty. It asks not only what happened to trade, wages or output, but how we know, whose experience the evidence represents and what alternatives were available at the time.
A larger factory, a rising export total or a new railway does not by itself prove that everyone became better off. Prices may have changed, working hours may have lengthened, records may exclude important groups, and the apparent turning point may have several causes. This guide shows how to examine those possibilities through source criticism, explicit comparisons and original worked examples.
The numerical cases below are fictional exercises, not estimates for real historical communities. They use invented places and generic currency units. Begin with the Gold Standard master, connect the method to Education: learning that transfers into life, or explore related systems through How X Works.
Choose your economic history route
Connect historical records to economic mechanisms, fair comparisons and defensible conclusions.
Route 1 · Chapters 1–3: Frame questions and read records
Route 2 · Chapters 4–6: Compare resources and technology
Route 3 · Chapters 7–9: Explain institutions and lived experience
Route 4 · Chapters 10–12: Trace change and test evidence
Route 5 · Chapters 13–14: Diagnose and practise
Chapter index
Frame questions and read records · Chapters 1–3
1. Turn a story about progress into a question you can investigate
2. Read a record as something somebody created for a purpose
3. Make historical units comparable before comparing quantities
Compare resources and technology · Chapters 4–6
4. Distinguish money earnings from purchasing power
5. Understand what historical output estimates can and cannot show
6. Explain technological change through complementary investments
Explain institutions and lived experience · Chapters 7–9
7. Trace institutions through rules and actual enforcement
8. Use counterfactuals without inventing a convenient alternative past
Trace change and test evidence · Chapters 10–12
10. Examine trade and structural change without double counting
Diagnose and practise · Chapters 13–14
13. Diagnose weak claims and practise changed-context reasoning
CHAPTER 1 OF 14 · Frame questions and read records
1. Turn a story about progress into a question you can investigate
Economic history begins when a broad narrative becomes a specific question. “The town prospered after the railway arrived” is an invitation to investigate. Did output increase? Did transport become cheaper? Did more people move in? Did landowners, workers and consumers experience the same changes? The answers may differ without contradicting one another.
Choose the outcome before searching for a convenient statistic. A question about household living standards needs evidence about resources and needs. A question about industrial organisation might need contracts, ownership records and production methods. A question about state capacity might require tax assessments, actual collections and administrative correspondence. Each source makes some activities visible and leaves others outside its frame.
Define the comparison
Suppose a fictional port handled 10,000 tonnes in one year and 15,000 ten years later. That establishes a 50% increase in recorded tonnage, if coverage and units are compatible. It does not establish a 50% increase in value, productivity or local income. Goods may be heavier but cheaper, ships may transship cargo without much local processing, or the reporting boundary may have expanded.
A good question might be: did the terminal reduce the resources needed to move a tonne between ship and warehouse? That requires labour hours, equipment use, handling losses and comparable cargo. Another question might ask whether waterfront households gained. That needs earnings, prices, employment security and perhaps housing costs. The same historical setting supports several distinct investigations.
The History root supplies the broader discipline of interpreting the past. The Economics root develops choices, incentives and resource constraints. Economic history brings them together while insisting that models fit the institutions and information of the period.
Do not begin with a verdict that a technology, empire or policy must have been beneficial. Begin with a claim small enough to test. A careful finding about one mechanism is more useful than a confident story whose evidence cannot distinguish between competing explanations.
CHAPTER 2 OF 14 · Frame questions and read records
2. Read a record as something somebody created for a purpose
A tax ledger is not a neutral census of everything people produced. A company balance sheet is not a complete account of community welfare. A newspaper advertisement records what a seller wanted readers to believe, which may differ from the terms buyers actually received. Before extracting numbers, identify who created the record, for whom, under what rules and with what incentives.
The US National Archives' written-document analysis guide provides a useful starting method: establish the document's form and origin, interpret it in context, and seek corroborating evidence. In economic history, add questions about the unit counted, reporting obligations and omissions. A perfectly transcribed number can still answer the wrong question.
Three documents, three different claims
Imagine an invented mill's prospectus promising annual output of 20,000 rolls, a manager's report recording 15,000 completed rolls, and a sales ledger showing 12,000 sold. None must be false. The first is a projection, the second a production measure, and the third a sales measure. Unsold inventory, defective goods, timing and reporting definitions can explain differences.
If a historian calls the prospectus figure actual production, the mistake occurs before any calculation. If the historian treats sales as production without examining inventory, another mistake follows. The repair is to label each variable and reconcile the flows rather than choosing whichever figure makes the strongest narrative.
Preserve the original transcription beside any cleaned version. Record uncertain handwriting, damaged pages, missing dates and editorial expansions. If a quantity is inferred from context, mark it as inferred. A reader should be able to retrace the path from the historical object to the analytical table.
Corroboration is not a vote
Five later books repeating one contemporary report do not provide five independent observations. Trace their citations. Conversely, two genuinely independent sources may disagree because they measure different populations or periods. Investigate their relationship before averaging them.
A useful source note states what the record directly establishes, what it only suggests, and what further evidence would resolve the uncertainty. This habit prevents archives from becoming collections of quotations selected to support a conclusion already decided.
CHAPTER 3 OF 14 · Frame questions and read records
3. Make historical units comparable before comparing quantities
Familiar words can conceal unfamiliar measures. A sack may be a container count, a customary volume or a standardised mass. A working day may vary by season and occupation. A monetary unit may change its metal content or relationship to other currencies. An apparent trend can therefore be a change in measurement rather than activity.
The existing Historical Metrology guide examines this problem through units, standards and provenance. For economic history, the practical sequence is to retain the original unit, establish a dated local definition, perform a justified conversion and preserve its uncertainty.
A grain shipment that appears unchanged
Suppose two fictional registers each record 100 baskets. In Harbour A, supporting evidence defines a basket as 20 litres; in Harbour B, it defines one as 25 litres. The shipments are 2,000 and 2,500 litres respectively. The identical recorded count conceals a 25% difference in volume.
Now suppose the first shipment cost 400 currency units and the second 450 in a compatible currency. Prices per basket are 4 and 4.5, suggesting a 12.5% increase. Prices per litre are 0.20 and 0.18, suggesting a 10% decrease. The conclusion reverses because the denominator has changed. Neither conversion establishes the grain's quality, moisture or nutritional content.
If the second basket's capacity is only known to lie between 24 and 26 litres, total volume lies between 2,400 and 2,600 litres. The price per litre lies between approximately 0.173 and 0.188. Reporting 0.180000 would conceal more uncertainty than it communicates precision.
Keep dimensions separate
Do not convert volume into mass without information about the substance and conditions. Do not interpret a daily wage as an annual income without the number of paid days. Do not compare total exports with domestic production when re-exports or imported inputs matter. Write the dimension beside each quantity before calculating.
When no defensible conversion exists, keep the original measure and narrow the claim. “The recorded number of baskets increased” can be justified even when “the mass of grain increased” cannot. Honest limits protect the usefulness of the evidence.
CHAPTER 4 OF 14 · Compare resources and technology
4. Distinguish money earnings from purchasing power
Nominal wages describe amounts of money. Real wages relate those amounts to prices. A wage can rise while the goods it buys decline. The US Bureau of Labor Statistics explains the general relationship in its income and CPI guide: adjusting nominal income by a suitable price index expresses purchasing power relative to a base period. Historical applications require particular care because the available prices and consumption patterns may be incomplete.
A wage increase that does not improve the selected comparison
In a fictional town, a worker's daily wage rises from 10 to 12 units. A fixed consumption basket rises from 5 to 6.5 units. The nominal wage increases by 20%, but basket purchasing power falls from 2 baskets per day to approximately 1.846. The decline is about 7.69%.
This result concerns the specified basket and daily wage. It does not yet measure annual household welfare. Suppose paid days rise from 200 to 240. Annual earnings rise from 2,000 to 2,880. Dividing by the basket prices gives 400 and approximately 443.08 basket units annually: an increase of about 10.77%. More days of work reverse the annual purchasing-power result, while also requiring more working time.
The historian should report both relationships. A household may obtain more goods over the year while the reward for each day's work falls. Whether its circumstances improved also depends on hours, risk, health, unpaid work and preferences.
Ask whose basket is represented
A basket dominated by grain may not represent an urban household facing sharply rising rent. A price series from a capital city may not represent remote villages. Employer-provided meals or housing may be part of compensation, but their value and quality need evidence.
Compare plausible baskets when the conclusion is sensitive to weights. Separate observed prices from reconstructed ones. Explain whether the series uses market prices, official prices or actual transaction evidence. Under rationing, a controlled price may not describe the cost of obtaining the quantity a household needs.
A real-wage series is a disciplined comparison, not a complete biography. Its strength comes from clearly defined boundaries and sensitivity checks, not from claiming to summarise every aspect of life in one number.
CHAPTER 5 OF 14 · Compare resources and technology
5. Understand what historical output estimates can and cannot show
Historical national accounts reconstruct economic activity from uneven surviving evidence. Researchers may combine agricultural quantities, industrial records, service estimates, prices and population figures. Earlier estimates can be less directly observed than modern accounts, although modern statistics also involve definitions and estimation. Always examine the method behind a smooth-looking series.
The University of Groningen's Maddison Project Database 2023 release provides long-run comparative income and growth estimates. Its documentation identifies the release, coverage and underlying research. Bolt and van Zanden's methodological paper explains why international purchasing-power benchmarks affect historical comparisons. Dataset versions and methods belong beside a chart, not hidden in an inaccessible footnote.
Total growth and growth per person
Consider a fictional economy whose real output increases from 100 million to 150 million units while population rises from 100,000 to 140,000. Total output grows by 50%. Output per person rises from 1,000 to approximately 1,071.43, an increase of about 7.14%. A claim about a 50% improvement in average resources would confuse aggregate and per-person changes.
If population at the second date is uncertain between 135,000 and 145,000, output per person lies between approximately 1,034.48 and 1,111.11, holding the output estimate fixed. If output is also uncertain, the range can widen. These are scenario bounds from stated assumptions, not statistical confidence intervals.
Output is not the same as distribution or welfare
An average does not reveal how income was shared. Production can increase while some people lose land, bargaining power or access to resources. It can also impose environmental damage or coercive working conditions that a narrow output measure does not value adequately. Add relevant evidence rather than demanding that one statistic answer every question.
For international comparisons, distinguish exchange-rate conversion from purchasing-power comparison. Neither should be selected simply because it produces a preferred ranking. Explain the purpose of the comparison, the benchmark and the time series used to connect dates. Apparent precision across centuries must not erase uncertainty in the foundations.
CHAPTER 6 OF 14 · Compare resources and technology
6. Explain technological change through complementary investments
A machine changes production only when it can be installed, maintained, supplied and used. Skills, energy, finance, standards and transport can be as important as the invention itself. Economic history therefore asks why adoption happened in one place and time, why it spread unevenly, and what institutions made it workable.
A fictional workshop decision
Suppose a workshop produces 1,000 identical units using 500 labour hours. A machine allows the same output with 300 direct labour hours but requires 40 maintenance hours. Labour productivity rises from 2 units per hour to approximately 2.94 when both direct and maintenance labour are included. Counting only direct labour would give 3.33 and overstate the improvement on this boundary.
Assume labour costs 5 currency units per hour. The old labour bill is 2,500. The new bill is 1,700. If annualised machine costs and additional energy total 600 for this production run, total measured costs fall from 2,500 to 2,300: a saving of 200. Labour-hour savings are not identical to total resource savings.
If output is only 500 units and the machine's fixed costs remain unchanged, profitability may differ. If reliable power is unavailable, the assumed operating pattern may be impossible. If skilled maintenance workers must be recruited from elsewhere, the historical cost and timing matter. Technology is an arrangement of complements rather than a self-executing object.
Adoption and consequences are separate questions
A firm's reason for adopting a technology may be to reduce cost or improve consistency. The consequences can include changes in skills, employment, consumer prices and bargaining relationships. Those outcomes need separate evidence. A machine's physical capability does not prove that employers passed savings to consumers or workers.
The eduKate article on Singapore's containerisation transition provides a practical reading route into connected equipment, land, labour and transport. Use it to ask what had to change around a new technology. Do not infer an entire national growth effect from a photograph of a crane or a terminal opening alone.
CHAPTER 7 OF 14 · Explain institutions and lived experience
7. Trace institutions through rules and actual enforcement
Institutions shape what people can own, promise, exchange and enforce. Formal rules may describe property, credit, taxation or work, while informal practices influence how those rules operate. A law's existence does not prove universal compliance, equal access or predictable enforcement.
Imagine two fictional markets with the same written contract rule. In one, disputes are resolved quickly at modest cost. In the other, resolution is slow and uncertain. Traders can respond differently even though the legal text looks similar. They may demand cash, restrict transactions to known partners or charge more for credit. The mechanism connects enforcement to behaviour rather than treating a legal label as sufficient explanation.
Follow a credit transaction
A merchant sells goods worth 100 units on credit. Under an invented model, repayment occurs with probability 0.9 and nothing is recovered otherwise. Ignoring all other costs, a promised repayment of approximately 111.11 has an expected value of 100. If repayment probability rises to 0.95, the corresponding amount falls to approximately 105.26.
These figures demonstrate a mechanism, not actual historical interest rates. Real contracts can include collateral, partial recovery, risk aversion, bargaining power, administrative costs and legal restrictions. A high observed rate might reflect several of these factors. It cannot automatically be interpreted as pure risk or pure exploitation without further evidence.
Distribution matters within an institutional arrangement
A rule can make transactions more predictable for one group while excluding another. Property records may recognise some claims and erase others. Labour contracts may be formally documented while workers have little practical freedom to leave. Economic analysis should identify whose options expanded and whose narrowed.
Avoid explanations that attribute centuries of outcomes to a single timeless institutional score. Establish the relevant mechanism and chronology. Look for changes in enforcement, access and behaviour, not only the date a rule was announced. Compare competing accounts, including whether economic change itself encouraged institutional reform.
Historical institutions are specific arrangements among people with unequal resources and power. Treating them carefully makes economic explanation more precise and keeps moral consequences visible without substituting moral judgement for evidence.
CHAPTER 8 OF 14 · Explain institutions and lived experience
8. Use counterfactuals without inventing a convenient alternative past
To claim that an event caused an outcome is to compare what happened with what would otherwise have happened. The alternative is unobserved, so it requires justification. Economic history uses comparisons, models and detailed contextual evidence to make that reasoning explicit. A counterfactual is useful when it clarifies assumptions; it is misleading when it is presented as another observed fact.
A transport-cost exercise
Suppose a fictional railway carries 10,000 tonnes of a commodity at a resource cost of 2 units per tonne. An available canal alternative would cost 5. For the same quantity, the simple resource-cost difference is 10,000 × (5 − 2) = 30,000 units. This is a restricted comparison under fixed quantity and route assumptions.
It does not establish the railway's entire contribution to national income. The calculation omits construction, induced demand, reliability, travel time, external effects and changes in settlement. It also assumes the canal could actually carry that traffic. If capacity expansion were necessary, its cost belongs in the alternative. If traffic would be lower without the railway, fixed-quantity arithmetic answers a different question from a full welfare comparison.
Ask whether the alternative was feasible
The relevant alternative is not necessarily “nothing.” People might improve roads, relocate production, use another port or consume different goods. But alternatives must fit the technology, geography, institutions and resources of the period. Giving historical actors modern capabilities makes the exercise easier and less informative.
Compare several plausible alternatives when evidence does not choose one decisively. Explain what remains stable across them. If the conclusion changes radically, that sensitivity is a finding: the causal claim depends on an assumption that needs more investigation.
Counterfactual reasoning also helps separate importance from necessity. A technology may have delivered substantial benefits without being the only conceivable path to development. Conversely, the existence of an alternative does not show that adopting it would have been cheap, fast or politically feasible. The historian's task is to evaluate those constraints rather than win an argument through an unrealistic comparison.
CHAPTER 9 OF 14 · Explain institutions and lived experience
9. Follow households and workers beyond the average wage
Economic life occurs through households, workplaces and communities, not only national totals. A wage record may describe a narrow group of paid workers while missing home production, informal trade, seasonal work and unpaid care. Household income may combine several earners and transfers. A change in the recorded workforce can alter an average even when no individual's wage changes.
A composition effect
Imagine a town with 80 workers paid 10 units a day and 20 skilled workers paid 20. The average is 12. Ten years later, suppose there are 50 in each category at the same respective wages. The average becomes 15, a 25% increase. Every category's wage is unchanged; the composition has shifted.
That shift may be economically important, perhaps reflecting training or industrial change. But it is not evidence that a typical worker in each occupation received a 25% raise. Report occupation-specific changes alongside the aggregate. If the records cover different employers or exclude casual workers in one year, even the composition interpretation needs qualification.
Time and risk belong in the account
Two workers earning the same annual amount may face different hours, injury risks, unemployment spells and commuting burdens. A household can increase money income by working longer while losing time for education or care. More income can still be valuable, but the historical comparison should not erase how it was obtained.
Look for family budgets, working-time records, housing evidence and accounts of everyday life. Such sources also have selection problems: a surviving household diary is not automatically representative. Use it to illuminate a mechanism or experience, then be clear about the limits of generalisation.
The Demographic Economics guide explains how age, work, consumption and transfers interact. Economic history adds another layer: definitions and institutions can change across generations. A category called dependent, employed or household may not mean the same thing in every historical census. Read the instructions behind the classification before comparing totals.
CHAPTER 10 OF 14 · Trace change and test evidence
10. Examine trade and structural change without double counting
Trade records can reveal connections among places, but gross flows do not equal local income. Imported inputs may be processed and exported. Goods may pass through a port more than once. Price changes can raise export values without increasing physical quantities. A country can gain from trade while some groups face losses or difficult transitions.
Gross exports and local value added
In an invented processing business, exports sell for 1,000 units. Imported materials cost 600 and domestically purchased intermediate inputs cost 150. The business's own value added is 250 before distributing it among labour, capital and relevant taxes. Calling all 1,000 domestic income would count inputs as though the business had produced them from nothing.
The domestic supplier's contribution must be considered separately if the question concerns the whole domestic production chain. Its sales of 150 are not automatically all value added either; it may use imported inputs. Define the boundary and trace intermediate transactions before aggregating.
A shift among sectors
Suppose 60% of workers are in a sector producing 10 units per worker and 40% in a sector producing 30. Average output per worker is 18. If shares change to 40% and 60% with sector productivity unchanged, the average becomes 22. The increase reflects reallocation under these assumptions, not improvement within either sector.
Actual structural change can involve both mechanisms. Workers may need training, housing and transport to move. The expanding sector may not absorb everyone at its previous productivity. Urban congestion, capital requirements and unequal access can alter the outcome. A decomposition organises the question; it does not prove that movement alone caused the gain.
Historical trade analysis should also examine terms, bargaining relationships and coercion where relevant. Export growth is not a sufficient welfare test. Ask who controlled production, who received payment, who bore risk, and what happened to access to land, food and other resources. These questions make the account more complete without assuming the same answer for every place and period.
CHAPTER 11 OF 14 · Trace change and test evidence
11. Treat crises as sequences, not single-date explanations
Wars, financial disruptions, harvest failures and epidemics can affect economies through several channels. A crisis label does not identify which channel mattered most. Establish the sequence: what changed first, what people knew, which markets or institutions were affected, and how responses altered the outcome.
For a fictional grain market, a poor harvest might reduce supply, transport disruption might prevent imports, and household income loss might reduce access to food. Rising prices alone cannot identify the balance of these mechanisms. Recorded supply can fall while some households retain purchasing power; aggregate supply can remain substantial while vulnerable groups cannot obtain enough.
A business failure is not necessarily a production failure
Imagine a workshop with positive operating revenue over a year but a large loan repayment due before customers pay their invoices. It may face a liquidity problem even if its long-run assets exceed liabilities. Conversely, repeated losses can indicate a deeper solvency problem. Historical accounts should distinguish payment timing from the underlying capacity to cover obligations.
Suppose cash available is 20 units, receipts due next month are 80, and a payment of 60 is due today. The immediate gap is 40. Selling an asset, obtaining short-term credit or delaying payment could change the outcome. Whether those options existed depends on contemporary institutions and relationships. The annual revenue total alone does not explain survival.
Avoid explaining everything with hindsight
An action that looks mistaken after the outcome may have been reasonable under contemporary information. Identify what decision-makers could observe and which risks they faced. At the same time, uncertainty does not excuse unsupported claims about their motives. Letters, minutes and policy documents may help, but each has a purpose and audience.
Compare recovery across sectors and groups. A return of aggregate output to its previous level does not mean households recovered lost assets or opportunities. A crisis can leave lasting changes in debt, health, education or migration. Follow those channels when making claims about persistence rather than assuming that a recovered total means an unchanged society.
CHAPTER 12 OF 14 · Trace change and test evidence
12. Build a reproducible historical comparison
A reader should be able to understand how a conclusion was produced. Preserve the source reference, transcription, cleaning decisions, conversion rules and calculations. Keep missing values distinct from zeros. Identify interpolated years and avoid drawing a continuous line that makes sparse observations appear annual and precise.
A simple before-and-after problem
Suppose output in a fictional district rises from 100 to 140 after a new road opens. A similar district without the road rises from 100 to 120. The raw increase in the first district is 40; subtracting the comparison increase gives 20. This difference-in-differences calculation is an organising device for a causal question, not automatic proof of a road effect.
The comparison needs a credible reason to represent what would have happened without the road. Perhaps the first district was already growing faster, attracted a new factory, or received other investment. Perhaps the road benefited the comparison district too. Examine earlier trends, concurrent changes and spillovers. A neat calculation cannot repair an unsuitable comparison.
Test the decisions that could change the conclusion
Recalculate using another plausible price index, population estimate or boundary definition. Separate results for subgroups when the source permits it. Ask whether excluding one unusually influential observation changes the story. These checks should address real weaknesses, not become a ritual list of statistical exercises.
State the level of confidence in ordinary language. “The evidence is consistent with lower handling costs” is different from “the terminal caused all subsequent growth.” Explain what additional records would strengthen or weaken the claim. This makes the analysis useful to someone who may later discover new material.
Reproducibility includes interpretation. A spreadsheet with formulas is insufficient if the variables' meanings are unclear. A narrative with footnotes is insufficient if numbers cannot be traced. The strongest account allows readers to inspect both the calculation and the historical reasoning that makes it relevant.
CHAPTER 13 OF 14 · Diagnose and practise
13. Diagnose weak claims and practise changed-context reasoning
When a historical claim seems too easy, identify the missing bridge. A nominal wage increase needs a price comparison. A production increase needs compatible coverage. A technology story needs complementary inputs. A causal claim needs an alternative explanation to be considered. A welfare claim needs distribution and human experience beyond the aggregate.
Repair a misleading headline
Consider this invented statement: “Mill workers became twice as prosperous because exports doubled.” The evidence supplied is export revenue rising from 100,000 to 200,000. What would you need before accepting the conclusion?
First distinguish prices from quantities. Next determine local value added, employment and compensation. Then examine worker earnings, paid days and household prices. Finally consider how gains were distributed. Export revenue may have doubled because prices rose, imported inputs increased, or ownership returns grew. None of those possibilities establishes a doubling of workers' purchasing power.
Changed-context exercise
Return to the daily-wage case. The wage rises from 10 to 12 units. This time the relevant basket rises from 5 to 5.5, while paid days fall from 200 to 180. Has daily purchasing power risen? Has annual purchasing power risen?
Daily basket purchasing power rises from 2 to approximately 2.182, an increase of about 9.09%. Annual earnings change from 2,000 to 2,160. Annual basket purchasing power changes from 400 to approximately 392.73, a fall of about 1.82%. The same worker can have a better daily real wage and a lower annual purchasing-power total.
Now suppose the worker receives meals previously purchased from earnings. The calculation needs the meals' quantity, quality and relevance to the basket. Do not add an arbitrary modern price. If the evidence is inadequate, state that the cash-only comparison omits an in-kind component rather than inventing a complete compensation estimate.
The transfer skill is to recognise the relationship when the setting changes. Whether the subject is a dockworker, farm household or factory employee, define the resource, price, time and population before interpreting the result. The question is what the evidence supports, not which calculation produces the most dramatic number.
CHAPTER 14 OF 14 · Diagnose and practise
14. Questions, learning routes and primary reference points
Is economic history only the history of money?
No. It includes production, work, consumption, technology, institutions, trade, finance, inequality and the use of resources. Money records are often valuable because they survive in accounts, but economic life also includes non-market activity and relationships that accounts may omit.
Can modern economics be used to study earlier societies?
Yes, when its assumptions are examined rather than imposed automatically. Resource constraints, incentives and accounting relationships can clarify questions. Historical evidence must establish whether the relevant markets, rights, information and choices existed. A model is an aid to explanation, not permission to replace the past with a simplified modern setting.
Are old statistics too unreliable to use?
Not necessarily. Their usefulness depends on provenance, coverage and the question. A partial ledger can establish a particular transaction even if it cannot represent an entire economy. Carefully reconstructed series can support broader comparisons, provided the method and uncertainty remain visible. Rejecting all imperfect evidence would also discard much of what can responsibly be learned.
Does economic growth mean historical living standards improved?
It can contribute to improvement, but the inference requires additional evidence. Distribution, prices, working time, health, security and access to resources matter. Compare the relevant outcomes and explain why they may move differently. Neither a rising average nor a vivid individual account should be asked to represent everything.
How should a beginner organise a small project?
Choose one bounded question, such as how a local transport change affected handling work. Assemble a small set of sources with different perspectives. Create a table of definitions and uncertainties. Make one transparent calculation, test an alternative interpretation, and write a conclusion that distinguishes observation from inference. A modest project done carefully teaches more than a sweeping chronology assembled from disconnected facts.
For source interpretation, continue with History. For measurement, use Historical Metrology. For technology and place, explore Singapore's container-port transition. For age and household support, use Demographic Economics.
Sources and their roles
The National Archives document-analysis resources support systematic reading of primary records. The BLS guide to income and prices explains the nominal–real distinction; its modern index is not a ready-made deflator for every historical society. The Maddison Project release page and Bolt–van Zanden methodological paper provide a route into historical national accounts and benchmark choices.
The worked cases are original illustrations, not extracted historical estimates. Their purpose is to expose assumptions so that readers can replace them with documented evidence. A dependable economic history ends with a conclusion proportionate to its sources, and leaves a clear path for the next researcher to check, challenge or improve it.