Introduction: The Essential Role of Long‑Run Wage and Price Data

The construction of consistent historical wage and price series is a foundational activity in empirical economics. These series—often stretching back centuries—provide the raw material for understanding how living standards, labor markets, and price levels have evolved. Without reliable historical data, economists cannot test theories of long‑run growth, evaluate the impact of monetary or fiscal policy, or calibrate models used for forecasting. This article examines the methods, challenges, and applications involved in developing such series, emphasizing why they remain an indispensable tool for economic modeling.

Why Historical Wage and Price Series Matter for Modeling

Economic models rely on parameter estimates that are generally derived from historical data. Wage and price series offer the time dimension that makes these estimates meaningful. They allow researchers to:

  • Separate cyclical fluctuations from secular trends.
  • Identify structural breaks caused by wars, depressions, or technological shifts.
  • Test theories of income distribution, inflation dynamics, and productivity growth.
  • Calibrate dynamic stochastic general equilibrium (DSGE) models and other macro frameworks.

For example, a model that seeks to explain why real wages have stagnated in certain economies while productivity rose must be grounded in wage and price data that are comparable across decades. Similarly, inflation targeting regimes depend on price indices that accurately capture historical behavior. The credibility of any such model rests on the quality of the underlying series, making careful development a prerequisite for sound analysis.

Linking Past and Present: The Need for Consistent Metric

A wage or price observation from 1800 is not directly comparable to one from 2025 because the goods consumed, the types of employment, and the methods of data collection differ radically. Historical series must therefore be linked through careful adjustments and splicing. A modern dollar does not buy the same basket as a dollar from 1920; the decennial changes in consumer preferences, product quality, and measurement standards require constant vigilance. The result is a set of series that, while imperfect, provides the best available representation of long‑term economic realities.

Core Challenges in Building Reliable Historical Series

Developing wage and price series that span more than a few decades presents a unique set of difficulties. These challenges can be grouped into data scarcity, changes in definitions, and the need for standardization.

Data Scarcity and Fragmentary Sources

Before the twentieth century, systematic collection of wages and prices was rare. Researchers must rely on fragmentary records: merchants’ ledgers, pay rolls from estates, newspaper advertisements, and government reports that were often ad hoc. In many countries, wage data exist only for specific occupations (e.g., agricultural laborers, carpenters) or regions. The challenge is to generalize from these scattered observations to a national average, using techniques such as:

  • Mediation through proxy variables – If wages for skilled workers are missing, researchers may use agricultural day‑labor rates as a substitute, adjusting for skill premiums.
  • Interpolation across census years – When data are available only every decade, linear or log‑linear interpolation fills the gaps.
  • Multiple source reconciliation – Comparing price lists from different areas helps identify systematic biases (e.g., urban vs. rural cost of living).

Even with these methods, early data carry wide margins of error. Acknowledging those uncertainties in model specifications is critical.

Changes in Definitions and Classification Systems

Occupational titles, industry classifications, and consumption baskets shift over time. A “weaver” in 1800 might work on a handloom; by 1900, the same term could refer to a factory machine operator. Similarly, the price of a “loaf of bread” changes in size, quality, and even type of bread. For price series, the typical consumer basket must be updated regularly to reflect new goods and changing tastes. Chain‑weighting methods—where each year’s index uses weights from a recent base period—help mitigate the substitution bias, but they require detailed expenditure data that may not exist for earlier periods.

Inflation Adjustments and the Real–Nominal Distinction

Real wages are nominal wages divided by a price index. Constructing a meaningful price index over long periods is itself a complex task. The index must: (a) include a representative basket of goods, (b) account for quality changes, (c) use consistent weights, and (d) reflect the cost of living for the target population (e.g., urban workers vs. farm households). Many historical wage series are published in nominal terms; users must apply a deflator, often from a separate source. Inconsistencies between the wage deflator and the consumer price index can lead to spurious trends. Researchers often prefer to reconstruct both wage and price series from the same underlying data to ensure coherence.

Methodological Approaches to Constructing Long‑Run Series

A variety of statistical and historical methods have been developed to overcome the data challenges. These approaches share a common goal: produce a consistent time series that can be used in econometric analysis.

Primary Source Digitization and Archival Research

Many historical databases are built by hand‑transcribing data from original records. The MeasuringWorth project, for example, has compiled thousands of wage and price observations for the United States and the United Kingdom, drawing on government publications, commercial almanacs, and academic studies. This painstaking work involves not only entering numbers but also interpreting units of account, currency reforms, and seasonal adjustments.

Standardization and Concordance Tables

When linking series from different sources, researchers create concordance tables that map occupational titles and commodity descriptions from one period to another. For instance, the U.S. Bureau of Labor Statistics (BLS) has published a Consumer Expenditure Survey since the 1880s, but the categories change over time. A modern researcher might assign “kerosene lamps” from an 1890 price list to a “lighting” category that later includes electric lighting, using explicit conversion factors for lumens and cost per hour.

Backcasting and Imputation

When data exist for the modern period but not for the distant past, economists may backcast using known relationships. For example, if real GDP and the wage share are known for certain benchmark years, one can impute missing wage series using a production function. Similarly, price series for specific goods may be imputed from related commodities (e.g., using rice prices to fill gaps for coarse grains when direct observations are missing). These imputations require strong assumptions and are usually accompanied by sensitivity analysis.

The Use of Purchasing Power Parity (PPP) for International Comparisons

Historical wage and price series are frequently used in cross‑country growth models. To compare real wages across nations, researchers must convert local currencies using PPP exchange rates rather than market exchange rates. Long‑run PPP databases, such as the ones provided by the Maddison Project or the OECD Historical Statistics, provide benchmark estimates that allow for meaningful comparisons. The process involves adjusting for differences in the composition of consumption baskets and for non‑traded goods—a notoriously difficult task for centuries past.

Applications in Economic Modeling

Historical wage and price series are not mere historical curiosities; they are actively used in modern economic analysis and policy design. Below are several key applications.

Estimation of Long‑Run Income Inequality

Studies of inequality rely heavily on wage series. The well‑known U‑shaped pattern of inequality in the United States (from the 1920s to the 1970s, then rising after 1980) is derived from labor‑income data that have been carefully standardized across decades. Without a consistent wage series, it would be impossible to compare the earnings of a factory worker in 1950 with one in 2020. Price series also matter: the inequality of real consumption depends on the prices faced by different income groups (e.g., the rich spend more on services, which have risen in price relative to goods). Historical series allow researchers to compute separate cost‑of‑living indices for different deciles.

Heating‑Up the Phillips Curve Debate

The Phillips curve—the relationship between unemployment and wage inflation—depends on accurate wage data. Historical series help economists test whether the curve has flattened or shifted over time. For instance, the apparent breakdown of the Phillips curve in the 1990s prompted a re‑examination of wage measurement: was it due to measurement error (e.g., changes in compensation forms such as bonuses and benefits) or a genuine change in economic structure? Long‑run wage series that include both base wages and total compensation are essential to answering this question.

Indexing Social Security and Pensions

Governments use historical price indices to adjust Social Security benefits, pension payments, and tax brackets for inflation. The choice of price index—Consumer Price Index for All Urban Consumers (CPI‑U) vs. CPI for the Elderly (CPI‑E), for example—can have large fiscal effects over time. Historical series are used to simulate the effects of alternative indexing formulas, such as the Chained CPI, which is designed to better reflect substitution behavior. The development of these series requires consistent back‑data to evaluate how different indices would have performed in the past.

Macroeconomic Model Calibration

Modern DSGE models require parameters for the steady‑state real wage, the markup of prices over marginal cost, and the elasticity of substitution between capital and labor. These parameters are often calibrated using historical averages from wage and price series. For instance, a calibration of the U.S. economy might target a historical average real wage growth of 2% per year and a wage share of labor of 60%. If the underlying wage series are not consistently defined, the calibration will be biased, leading to poor model performance.

Case Study: The U.S. Bureau of Labor Statistics Historical Wage Series

The BLS has been collecting wage data since 1884. Its “Industry Wage Survey” program began in 1907, and the “Employment, Hours, and Earnings” series dates from 1909. These data have been painstakingly revised over time to incorporate changes in industrial classification (e.g., from SIC to NAICS) and to adjust for shifts in the composition of the labor force. The BLS also maintains a historical price index that goes back to 1913 (the start of the CPI). Researchers can download these series for free from the BLS databases and link them with modern series using the methods described above.

A major challenge with BLS data is the break in 1940 when the agency changed from a “money wages” concept to “real wages” derived from cost‑of‑living data. For earlier decades, only nominal wages are available, requiring the researcher to apply a deflator. The BLS provides its own historical CPI, but the two series were not necessarily constructed on the same sampling frame. Modern work often reconstructs a harmonized series.

Case Study: Historical Wage Series in the United Kingdom

The United Kingdom has one of the longest continuous wage series in the world, thanks to the work of economic historians such as E.H. Phelps Brown, Gregory Clark, and Robert Allen. Using data from estate records, church accounts, and government surveys, researchers have built wage series for agricultural and building workers going back to the 13th century. The UK Office for National Statistics (ONS) now publishes annual data on average weekly earnings from 1963 onward, but for earlier periods one must rely on academic databases such as the “History of Wages in Great Britain” project. These series are used to study the Industrial Revolution, the wage gap between men and women, and the evolution of living standards.

Future Directions: Integrating Big Data and Digital Sources

The development of historical wage and price series is not a finished task. In the 21st century, economists are exploring new sources such as digitized newspapers, online job advertisements, and scanner data from retail stores. These data streams offer higher frequency and greater granularity but also introduce new biases (e.g., only covering internet‑connected populations). The challenge is to link these modern digital series with the carefully reconstructed historical series to create seamless long‑run databases. Machine learning techniques can help to harmonize classification systems and impute missing values, but the fundamental principles—documentation, comparability, and transparency—remain unchanged.

Conclusion

Historical wage and price series are the bedrock of quantitative economic history and macroeconomic modeling. Constructing them requires a blend of archival detective work, statistical skill, and sound theoretical judgment. While the data are always imperfect, the process of building these series forces researchers to be explicit about their assumptions, which in turn makes economic models more robust. Whether the goal is to understand the causes of the Great Depression, to calibrate a growth model, or to design a cost‑of‑living index for pensioners, the availability of high‑quality historical series is essential. As new data sources emerge and computational methods improve, the potential for even richer historical databases grows. The careful development of these series will remain a vital endeavor for the economics profession.