Unbiased Analysis of Today's Healthcare Issues

Tax Collection in History

Written By: Jason Shafrin - Sep• 05•06

On Thursday I mentioned the Treasury Department’s plan to turn over some of the collection of delinquent taxes to private companies.  In most modern societies of the Western world, tax collectors are salaried employees of the state.  This was not always the case.  A 2006 paper by Cosgel and Miceli aims to create a model as to why governments choose the tax collection systems they do.  They also examine the tax collection methods of the Ottoman empire to help support their thesis.  The authors claim that the agency relationship between the government and the tax collectors is the determining factor as to which system is chosen.  The three competing systems are 1) share contract, 2) rent contract and 3) wage contract.  I will discuss each in detail below.

Share Contract

The share contract occurs when the government grants the agent a fixed percentage of all tax revenue received.  This is the system which is currently proposed by the U.S. government.  Share contracts were also used in pre-revolutionary France, India, China and Medieval Egypt.  The model Cosgel and Miceli employ is one of a tax collector attempting to maximize utility by choosing an effort level ‘e‘:

  • U=E[f(e)B-T(f(e)B)]-e

The tax base is represented by ‘B‘ and ‘f(e)‘ is a stochastic function.  The compensation to the tax collector ‘T()‘ is a function of the total revenue collected.  The first order condition for this model is:

  • f'(e)E[B(1-T’)]=1

Once the collector chooses the optimal e*, the value of the tax collection to the government is:

  • f(e*)E[B]-e*-k-u

where ‘u‘ is the reservation utility and ‘k‘ is the fixed cost to measuring the tax.

Rent Contract

In a rent contract the collector would pay the government a fixed amount and in turn would receive the right to all taxes of the specified area in return.  Republican Rome, the Abbasid Empire and the Ottomans all heavily used rent contracts.  Sometimes these contracts were given to military officers in lieu of wage payments for their services.  The objective function and first order conditions for the tax collector are below:

  • U=f(e)B-e-T(B)
  • f'(e)B=1

The rent contract is the only one in which effort is state contingent [e* is a function of the stochastic variable ‘B’], because the tax collector’s payment to the government is determined ex ante but the actual revenue ‘R‘ is determined ex post

The return for the government is:

  • V=E[f(e*(B))B-e*(B)]-m-u

Here, effort is a function of the tax base since more effort will be put forth when the tax base is larger.  The cost of the government verifying the tax base is ‘m’, however if there is competitive bidding for the contract, then m=0.

Wage Contract

Wage or salaried workers are the most common form of compensation for tax collectors today.  Eighteenth century Sweden and Russia used salaried agents to collect taxes.  After tax collection cartels in England (The Great Farm) and France (Ferme Générale) reduced the percentage of tax revenue the governments were receiving, both countries switched to a salaried system in the mid to late 18th century.  The objective function and FOC for this model are below:

  • U=f(e)E[B]-se-w(e),    s.t.: w(e)-e>u
  • f'(e)E[B]=1+s

Here ‘s‘ is the cost of monitoring.  In this model, the government receives:

  • V=f(e*(s))E[B]-e*(s)(1+s)-u


The main determinant of which method to choose is to see whether it is more costly 1) to verify tax revenue, 2) to verify the tax base, or 3) to verify worker effort.  The three options correspond to a high level of 1) k 2) m  and 3) s.  Share contracts are not very common since it is often very difficult to determine how much tax revenue has been received.  The tax collectors could hide the revenue.  Verifying the tax base is often difficult in rural areas where farmers produce agricultural products.  Valuing crops for tax purposes is often difficult. 

The authors claim that the Ottoman empire used rent contracts for longer than did England or France because the tax base was easily calculated prior to the 19th century.  Surveys were conducted every twenty to thirty years and would remain relatively accurate.  At the beginning of the 20th century, the authors claim that the tax base became more variable due to “demographic and socio-economic transformations” yet the authors do not specify which transformations these are.  Nevertheless, when the Ottomans moved to a salaried system of bureaucrats in the late 19th century, the authors claim that these “transformations” (which increased the variability of the tax base) were the cause of this institutional change.

One point which is not emphasized in the paper is that in a competitive bidding process the rent contract should be the most efficient.  If there is competitive bidding, the government will not have to value the tax base, but can simply auction off the contract and bidders will increase the price until the contract price reaches the estimated tax receipts less labor costs for collecting the taxes.  It is possible that this is theoretically true, but has not been shown empirically throughout history (for instance due to cartels); yet the authors do not make this point and simply assume that the government must incur a cost to verify the tax base.  Overall, the paper provides a nice unifying principal-agent framework to the systems of tax collection which could be chosen by a government, but only provide weak empirical evidence that the reason a tax collection system was chosen in the past came from the incentives derived from their models.  The authors also skip over the problem of who monitors the monitors within the government. 

Cosgel, Metin; Miseli, Thomas; (2006) “Tax collection in history: Public institutions and institutional change in the Ottoman Empire“, Presentation at the International Economic History Congress, Helsinki, Finland; August 22, 2006. 

You can follow any responses to this entry through the RSS 2.0 feed. Responses are currently closed, but you can trackback from your own site.