Historians studying inequality face a fundamental empirical problem: the wealthy rarely volunteer information about their fortunes, and the poor leave even fainter documentary traces. Yet across centuries of European and colonial administration, one institution compelled disclosure with remarkable persistence—the tax assessor. Fiscal records, however imperfect, constitute our richest quantitative window into pre-modern wealth distributions.

The methodological challenge is substantial. Tax documents were created not to measure inequality but to extract revenue, and their structure reflects fiscal priorities rather than statistical ones. Underassessment, exemptions for nobility and clergy, evasion, and shifting tax bases all introduce systematic biases that must be identified and, where possible, corrected.

What emerges from careful reconstruction is striking. Recent work by Alfani, Piketty, Lindert, and others has produced wealth distribution series stretching back to the fourteenth century in some Italian and Dutch cities. These series reveal that inequality is neither a recent phenomenon nor a monotonic trend, but a variable that responded to demographic shocks, institutional changes, and economic transformations in ways that complicate simple narratives. The numbers tell a story that no contemporary chronicler could have perceived—and that requires rigorous quantitative treatment to extract.

Converting a tax assessment into a wealth estimate requires understanding what the assessor was actually measuring. A fifteenth-century Florentine catasto entry, an English Hearth Tax return, and a nineteenth-century Prussian income tax declaration each operate under distinct assumptions about what constitutes taxable capacity. The first step in any reconstruction is reverse-engineering the assessment formula.

Underassessment is the most persistent challenge. Comparative studies of Italian estimi against post-mortem inventories typically show assessed values at 30-60% of market values, with the ratio often declining as one moves up the wealth distribution—wealthier households possessed greater means to negotiate favorable assessments. Correction factors must therefore be wealth-dependent, not uniform, lest the reconstructed distribution understate top-end concentration.

Exemptions present a more intractable problem. Where ecclesiastical property, noble estates, or specific occupational categories fell outside the tax net entirely, the resulting distribution is truncated rather than merely scaled. Reconstruction requires external sources—monastic cartularies, feudal surveys, sumptuary records—to estimate the missing mass. Florence's exclusion of clerical wealth, for instance, removes perhaps 20-25% of the total stock from view.

Changing tax bases over time create comparability problems that can manufacture spurious trends. A shift from real property to comprehensive wealth taxation will appear as rising inequality even when the underlying distribution is stable, because previously invisible movable wealth—disproportionately held by merchants and financiers—suddenly enters the data. Splicing series across such transitions demands explicit modeling of the coverage change.

The methodological discipline this imposes is salutary. Each reconstructed series must be accompanied by an audit of its assumptions: which households are observed, at what fraction of true wealth, with what error structure. Without this transparency, quantitative claims about historical inequality reduce to numerical assertion.

Takeaway

A tax record is not a measurement of wealth—it is a measurement of what the state could see and chose to tax. The distinction governs every inference that follows.

The Gini coefficient dominates historical inequality measurement, and for defensible reasons: it is scale-invariant, bounded between 0 and 1, and computable from grouped data when individual observations are missing. But its convenience has bred overreliance, and serious quantitative work increasingly supplements it with distribution-sensitive alternatives.

The Gini's central weakness is its insensitivity to where in the distribution change occurs. A redistribution from the 90th to the 50th percentile and an equal-magnitude redistribution from the 99th to the 90th can produce identical Gini movements while representing economically very different events. For pre-industrial societies, where top-end concentration drives most variation, this matters enormously.

Top wealth shares—the fraction held by the top 1%, 5%, or 10%—have therefore gained prominence. They are directly interpretable, comparable across studies, and focus attention on the segment of the distribution where fiscal records are typically most reliable (the wealthy being harder to omit entirely, even if underassessed). Alfani's work on Italian cities relies heavily on top-decile shares precisely because they prove robust to many of the biases that distort full-distribution Ginis.

Theil indices and generalized entropy measures offer further analytical leverage, particularly their decomposability into within-group and between-group components. This permits questions that Ginis cannot answer: how much of overall inequality reflects urban-rural gaps, or differences between guild members and outsiders, versus inequality within these groups? For historical analysis, where group structure often carries causal weight, decomposability is not a luxury.

Statistical significance and confidence intervals deserve more attention than they typically receive. Historical inequality estimates carry substantial uncertainty—from sampling, from correction factors, from coverage gaps—and point estimates without error bands invite overinterpretation. A movement in the Gini from 0.62 to 0.65 may or may not be meaningful depending on the underlying noise.

Takeaway

Choose your inequality measure to match the question you are asking. The Gini answers some questions well, others poorly, and presenting it alone often obscures more than it reveals.

The reconstructed series, taken together, dismantle the comforting view of inequality as a recent industrial pathology. Italian cities in 1300 already exhibited Gini coefficients above 0.7, with top-decile shares exceeding 60%. Pre-modern Europe was not an egalitarian peasant idyll disrupted by capitalism; it was a deeply stratified society whose inequality industrialization modified rather than created.

The Black Death of 1347-1351 produced the clearest exogenous shock in the historical record, and the data show its effect unambiguously. Across northern Italy, the Low Countries, and England, wealth concentration declined for roughly a century afterward, as labor scarcity raised real wages and pestilence redistributed property through inheritance disruption. By 1450, top-decile shares had fallen by 10-15 percentage points in several documented cities.

From approximately 1500 onward, however, inequality resumed a near-monotonic rise that persisted until World War I. This four-century trend—visible in Piedmont, Tuscany, Holland, and elsewhere—coincided with commercialization, state-building, and the early phases of globalization, but its drivers remain contested. Alfani emphasizes regressive fiscal systems and elite political capture; others stress factor-price movements and skill premia.

The twentieth-century compression of 1914-1980 appears, in this long view, as a historical anomaly rather than a culmination of progressive forces. Two world wars, depression, and the policy regimes they produced reduced top wealth shares to levels not seen since the medieval demographic collapse. Whether the post-1980 reversal represents reversion to a long-run mean or a distinct phenomenon remains an active research question.

These patterns challenge any monocausal theory. Inequality has responded to mortality shocks, war, institutional change, and economic transformation in ways that no single mechanism—neither Piketty's r>g nor Kuznets's industrialization curve—fully captures.

Takeaway

Inequality is not a one-way street. The historical record shows it can fall sharply when conditions change, but only rarely, and usually through catastrophe rather than reform.

Fiscal records, despite their distortions, have transformed our understanding of long-run inequality. What was once the domain of speculation and ideology has become an empirical field with explicit methods, contested findings, and accumulating evidence. The numbers do not speak for themselves, but they can be made to speak with discipline.

Significant gaps remain. Coverage outside western Europe is thin before the nineteenth century, women's wealth is systematically obscured by coverture and household-level assessment, and the integration of income and wealth measures across centuries is methodologically unsettled. Each represents an active research frontier.

The broader lesson is methodological. Historical questions about distribution, mobility, and structural change are answerable only through quantitative reconstruction, and quantitative reconstruction is credible only when its assumptions are transparent. The tax assessor, dead these many centuries, remains our most valuable informant—provided we understand what he was, and was not, recording.