Interactive companion

Tax Farming and the Origins of State Capacity in England and France

Noel D. Johnson and Mark Koyama · Explorations in Economic History, 2014 · Read the paper · Research

1Argument

Early modern rulers faced two costs. Collecting taxes was expensive in economies that were fragmented and hard to monitor, so rulers sold the right to collect a tax to a private farmer, who paid a fixed lease and kept whatever he collected. Borrowing was expensive too, because a king could not credibly promise to repay, so rulers depended on loans from the people closest to their revenues, including their tax farmers.

The paper builds a model that connects the two. When the crown does not need to borrow, competitive auctions are the best way to allocate tax farms, because bidding reveals which farmer can collect at lowest cost. When the crown needs large loans, competition fails, because a crown can default on any single farmer and re-lease the farm to another. A small group of financiers holding all the farms, a cabal, can punish a default together, so it can lend far more. Both England and France moved from competitive farming to cabal farming as their borrowing grew after 1600.

Cabal farming gave the crown a reason to invest in standardizing laws, taxes, and weights and measures, since it no longer learned collection costs from auctions. That investment undermined the cabal, because a more uniform fiscal system made direct collection a better outside option and lowered the cost of defaulting on the farmers. England, which was more uniform to begin with, abandoned its Customs farm in 1671 and its Excise farm in 1683. France kept its General Farms until the Revolution.

2The model

The paper is an analytic narrative. A model of how a crown allocates the right to collect its taxes yields four propositions, and the sections after this one read the English and French record against them. The model has a crown and many tax farmers and runs in three steps. The crown chooses how to allocate the farm. It chooses whether to invest in standardizing the fiscal system. It then borrows against the farm and decides whether to repay.

The controls in this section are linked. Moving a slider in one figure moves the same slider in every other figure.

Collection costs and the value of a farm

Each farmer has a private cost of collecting the tax. Costs are spread uniformly between a lowest cost θ̲ and a highest cost θ̄ with mean μ and standard deviation σ. Both depend on the country's underlying fiscal heterogeneity, which the paper ties to geography, and on whether the crown has invested in standardization. A farmer with cost θ values the farm at V(θ). Value falls as cost rises and falls faster at higher costs.

Three points on the distribution carry the model. The lowest-cost farmer has cost θ̲. A cabal is assumed to collect at the average cost, θ̂ = μ. A crown that collects directly faces the cost of the worst farmer, θ̄. With a uniform distribution θ̲ = μ − √3σ and θ̄ = μ + √3σ, so the spread of costs sets how far apart the three values lie.

value to the lowest-cost farmer, V(θ̲)
value to a cabal, V(θ̂)
value to the crown collecting directly, V(θ̄)
Figure 1. The value of the farm against the cost of collection. The shaded band is the range of farmers' costs. The paper assumes only that V falls and is concave. This page uses V(θ) = 4 − θ² with a mean cost of 1, and standardization halves σ and lowers μ by a tenth. Every number in this section is in those illustrative units.

Allocating the farm when the crown does not borrow

Under competition the crown sells the farm in an ascending-bid auction. The lowest-cost farmer wins and pays the value of the second-lowest-cost farmer. As the number of bidders grows the price approaches V(θ̲) and the crown captures nearly all the gains from contracting out collection.

Under a cabal the crown leases the farm to a single syndicate and the two split the surplus by Nash bargaining. The crown's outside option is direct collection, worth V(θ̄). The crown receives ½[V(θ̂) + V(θ̄)] and the cabal receives ½[V(θ̂) − V(θ̄)]. The crown gets more from a competitive auction than from a cabal. This is Proposition 1. A crown that does not need to borrow farms its taxes competitively.

Figure 2. What the crown receives under each allocation. The auction bar is the expected value of the second-lowest-cost farmer among n bidders, which is the price in an ascending auction. The dashed line is V(θ̲), the limit as bidders multiply. The cabal bar splits the farm's value under the cabal into the crown's share and the cabal's share.

Why a cabal makes standardization worth more

Standardization means common weights and measures, a more uniform tax code, and fewer openings for evasion. In the model it lowers both the mean and the spread of collection costs at a fixed cost. A lower mean helps the crown under every allocation. A narrower spread helps the crown when it collects directly or bargains with a cabal, because both payoffs rest on the high-cost end of the distribution. Under competition a narrower spread hurts the crown. The auction already pays it the value of the best farmer, and standardization erodes the advantage that farmer drew from local and tacit knowledge.

The crown therefore gains least from standardizing when it farms competitively. This is Proposition 2. An auction already reveals collection costs, so markets and standardization substitute for one another.

crown's gain from standardizing under competition
under a cabal
under direct collection
Figure 3. The crown's revenue under each allocation as the spread of costs changes, with the mean cost held at 1. The open circles mark the chosen spread. The filled circles mark the same crown after standardizing, which halves the spread and also lowers the mean cost, so they sit off the curves. The readouts give the gain from standardizing under each allocation. Whether the crown invests also depends on the cost of standardization, which the figure leaves out.

Borrowing and the commitment problem

Now the crown borrows an amount x against the second period of the lease. These advances are the anticipations of the sources. At the start of the second period the crown chooses whether to repay. Under competition a crown that defaults keeps the loan and re-leases the farm to one of the other bidders. Its only loss is the gap ΔB between what the original farmer would have paid and what the next bidder offers. The crown defaults whenever x ≥ ΔB. As bidders multiply ΔB shrinks toward zero, and a crown facing a perfectly competitive market defaults on any loan at all.

The paper treats the relationship as infinitely repeated and draws on two approaches to sovereign debt. Bulow and Rogoff (1989) show that many separate lenders cannot sustain lending without outside sanctions. Kletzer and Wright (2000) show that lenders who collude to punish a default together can. A cabal can punish together. If the crown defaults the cabal stops farming and the crown is left with direct collection in every later period. The crown repays when its discount factor δ satisfies δ ≥ (x − ΔV)/(x + ΔV), where ΔV = [V(θ̂) − V(θ̄)]/4. The same condition with ΔB in place of ΔV governs competition. A cabal's punishment is larger than the gap between two bidders, so a cabal can carry more lending. This is Proposition 3.

largest loan the crown repays under competition
largest loan the crown repays under a cabal
punishment for default, ΔB under competition against ΔV under a cabal
Figure 4. The lowest discount factor at which the crown repays a loan of size x, under competition and under a cabal. A loan is self-enforcing where its curve lies below the crown's δ, the dashed line. The competitive punishment ΔB is the expected gap in value between the lowest-cost and second-lowest-cost of n bidders, as defined in the paper's appendix. More bidders raise the competitive curve toward 1 at every loan size.

What wears a cabal down

The repayment condition does not rule out default. Default becomes likely when a war or a threat of revolution makes the crown weigh the present more heavily, which lowers δ. It also becomes likely when something narrows the gap between V(θ̂) and V(θ̄).

Standardization narrows that gap. So does lower underlying heterogeneity. Either lowers ΔV and with it the largest loan a cabal can sustain, and as σ approaches zero no lending survives at any δ below 1. This is Proposition 4. A cabal is more stable in a fragmented fiscal system than in a uniform one.

most the cabal can lend
most competitive farmers can lend at the same δ
Figure 5. The most a cabal can lend at each spread of costs, ΔV(1 + δ)/(1 − δ). Below the curve lending is self-enforcing and above it the crown prefers to default. The dot is the crown. With the default settings a loan of 2.5 is self-enforcing. Standardizing moves the dot left to the narrower spread and lowers the curve slightly through the lower mean cost, and the crown then defaults on the same loan. The faint curve is the curve before standardizing.

Propositions 2 and 4 together make cabal farming an intermediate institution. A crown that needs credit turns to a cabal. The cabal gives the crown a reason to standardize. Standardization shrinks the punishment the cabal can impose, and the crown's promise to repay stops being credible. The paper argues that the farmers could foresee this, and in both countries they tried to protect their position through ties to political elites.

Borrowing needs rise
→
Cabal farmingProposition 3
→
The crown standardizesProposition 2
→
The spread of costs narrows
→
The cabal cannot be repaidProposition 4
→
Direct collection

Risk aversion as an alternative explanation

A common explanation for tax farming is that a risk-averse crown traded expected revenue for a fixed payment. The paper gives three reasons this cannot carry the story. It says nothing about whether farming is competitive or centralized. In both countries farmers were granted defalcations, reductions in the lease after wars and other shocks, so the crown's receipts were far from fixed. And nothing suggests the English crown's attitude to risk changed in 1604, 1671, or 1683.

3Two paths through the model

The paper's Figure 3 places England and France on two axes, standardization and inside finance, between 1500 and 1800. Both began with competitive farming and low borrowing. Both moved up into cabal farming as borrowing grew. England started more standardized, so its cabal became unstable sooner and it moved to direct management after 1670. France stayed in the cabal region until the Revolution.

Figure 6. The four equilibria and the paths of England and France, redrawn from Figure 3 of the paper. The figure is a schematic, so positions are qualitative. The dots move between the dated points the paper marks, which are 1500, 1600, 1700, and 1800 for France and 1500, 1632, 1670, and 1800 for England.

4Lending through the French farms

No French data on lease prices or loans through the farms survive for the sixteenth century. For 1600 to 1656 the records kept by Jean-Roland Malet, published by Richard and Margaret Bonney, give the lease price of each farm and incomplete, and therefore lower-bound, figures on the loans made through them.

Real lease prices rose through 1643 and fell after. Loans through the farms ran from about half the lease price to more than its full value. At least before 1640 the high ratios came from more lending and not from falling leases. Between 1598 and 1655 only a third of leases ran their full contracted length, which is what the model predicts when a crown borrows from competitively allocated farms. The farms were consolidated in two waves, the salt, customs, and sales tax farms between 1578 and 1604, and then under Colbert into the single Company of General Farms by 1681.

5The Great Farm of the Customs

Elizabeth I put the Customs out to competitive farm from the late 1560s. James I was in one historian's words a super-borrower whose demands the London money market could not meet. In 1604 the Great Customs were consolidated into a single farm held by a syndicate that became the leading lender to James I and Charles I.

The rent of the first farm was £112,400. The crown did not push the rent up to what the farm could bear, and Ashton reads the low rent as the price the crown chose to pay for the new services the farmers performed. Under Charles I the crown began anticipating the rent several years ahead, and anticipations rose to almost half the annual rent by the mid-1630s. In 1641 Parliament expropriated the farmers. They were restored in 1660, but in 1671 the Customs farm was cancelled and the Customs placed under government supervision. The Excise followed in 1683. Both changes came before the Glorious Revolution.

6Standardization in France

To measure investment in standardization the paper counts the royal arrêts and edicts that limited five seigneurial institutions. These are the terrier, or seigneurial land register, seigneurial justice, the péages, or local tolls, corvée labor, and weights and measures. The list comes from Renauldon's 1765 treatise on seigneurial rights, and each rule change in an edict counts once.

Figure 9. Standardizing arrêts and edicts issued by the French monarchy by quarter century, 1500–1749, as in Figure 8 of the paper. The counts come from the authors' data file. The published figure shows 15 for 1725–1749, one more than the file holds.

The two jumps match the two waves of consolidation in the farms. The first follows the failure of the Grand Parti de Lyon in 1559, which ended the crown's use of outside finance. The second follows 1661, when Colbert was uniting the farms into the General Farms. Pick an institution to see each arrêt.

Figure 10. Each dot is a year in which the crown issued an arrêt or edict on that institution, from Table A1 of the paper. Hover a dot for the rule. Entries before 1500, such as the 864 ordinance of Charles the Bald on measures, are listed below the timeline but not drawn.

7Chronologies

Figure 11. Key dates in the move from competitive farming to cabal farming and, in England, to direct collection, from Tables 1 and 2 of the paper. Hover a mark for the event.

8Limits

The French loan figures are incomplete, so the ratio of loans to lease price is a lower bound, and no French series exists before 1600.

The paper's Figure 3 is a schematic. The positions of England and France on it are qualitative and the figures in Section 2 use one illustrative form of the value function.

The arrêt counts come from one treatise written by a seigneur. Renauldon wrote as an aid to other seigneurs facing royal encroachment, so the paper argues his list should if anything be biased against rulings that knocked down traditional rights.

The model explains why cabal farming was less stable in England. The paper also names the contingencies of the 1660s and 1670s, including a quarrel between the Lord Treasurer and the head of the farming syndicate, the costs of the Dutch War, and the king's spending.