Professional Documents
Culture Documents
Winter 2023-24
Introduction
• Thinking about organisations extends the definition of behavioural economics to include how
socialisation, networks, and identity shape individual behaviour in organisations
Introduction
• Systematic biases in individuals’ behaviour in firms – if workers misallocate their human capital –
can have large and persistent impact on decision-making in organisations
• How should organisations be designed to exploit – or correct – mistakes that stem from these
biases?
• A lot of psychology is involved when workers team up in organisations:
social comparison, changes in identity, attribution and diffusion of credit and blame,…
• (For more on this, see Camerer, Colin F., and Ulrike Malmendier, 2007. "Behavioral Organizational Economics." In Peter
Diamond and Hannu Vartiainen, eds., Behavioral Economics and Its Applications. Princeton University Press. Pages 235-
290)
Introduction
How to pick the ‘right’ people from a pool of applicants (screening, references,
interviews):
Assembling an applicant pool:
• where (how broadly) to search
• how long to search (vacancy duration)
• designing job ads and/or application forms
• how much to use networks and referrals (informal search)
Retention:
• setting the level of pay, benefits, duties, training
• creating an employer reputation
• structuring pay/benefits/duties
• offer matching policy
• evaluation, dismissal, layoff, leave policy
Personnel Economics is almost never about maximising profits at workers’ expense, for at
least two reasons:
1.
Good personnel policies in this module are defined as ‘win-win’ policies that make the pie bigger
→ We will try to identify policies that can make both firms and workers better off
→ ‘Good’ policies are Pareto-improving, Pareto-optimal, surplus-maximising, and socially efficient
Personnel Economics is almost never about maximising profits at workers’ expense, for at
least two reasons:
2.
Because there is a labour market, even an employer who cares only about their own profits cannot
treat workers arbitrarily badly. Why?
→ In many cases the fact that workers have outside options can force even purely selfish employers to
design Pareto-optimal personnel policies
Incorporating the constraints played by labour markets into the study of HRM is a key contribution
of personnel economics, relative to some other disciplines’ approaches
1. The exchange of effort for pay is probably the most important economic transaction in the
economy, and certainly in most peoples’ lives
2. The best way to structure these transactions is not at all obvious. We’ll show that two extreme
views:
• workers are inherently lazy and stronger financial incentives are always better than weaker ones
• workers are inherently good and will do what is needed with a minimum of financial incentives
can both lead to really bad results
3. Effective Human Resource Management (HRM) can be highly profitable, and can make workers’
lives much better too:
We’ll look at a number of case studies which show that simple, correctly-targeted HRM
innovations can yield gains that compare very favourably with the best technological innovations
(In fact, the two often go hand-in-hand)
• To motivate this model, imagine you have just been seriously injured in an accident in one of the
big box stores. You need to hire a lawyer to sue the store for damages.
• How should you pay this person?
(Figure 1.1, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
Our Example:
• One principal, one agent, no uncertainty
• One output (𝑄), observed by principal and agent. 𝑄 is dollars of net revenue.
• Principal can’t observe effort (𝐸) (so we can’t base the contract directly on it).
• The production function: 𝑄 = 𝑑𝐸
𝑑 > 0 is a productivity parameter that can capture ability or technology differences
Our Example:
• Much of the time, we’ll use our baseline production function :
𝑄 = 𝐸
When we use this function, we are measuring effort in terms of the amount of output it yields,
i.e. one unit of effort is ‘what it takes’ to produce one unit of output.
• The utility function (1) can be represented by an indifference map, which looks like this:
(Figure 1.3, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
• In general, the contract between the principal and agent can be any relationship between what the
principal observes (in this case 𝑄) and what the agent is paid (𝑌).
• In today’s example, we will restrict our attention to linear piece rate contracts,
i.e. to contracts of the form 𝑌 = 𝑎 + 𝑏𝑄:
→ The “contract” is an ordered pair, (𝑎, 𝑏), where 𝑎 is base pay and 𝑏 is the “piece rate” (again, per
dollar of net revenue generated by the agent’s effort).
(From Instructors’ resources, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
𝑈 ≥ 𝑈 𝑎𝑙𝑡
• (Assuming a linear contract (𝑎, 𝑏), and using our baseline production- and cost-of-effort functions)
and to 𝑄 = 𝐸
• Substituting the two constraints (and the baseline effort-cost function) into the maximand, this is
equivalent to:
max 𝑈 = 𝑎 + 𝑏𝐸 − 𝐸 2 /2
𝐸
• Solving this for the effort level the agent will choose, yields:
𝐸∗ = 𝑏
• So, a higher commission rate (𝑏) induces higher effort.
Changing base pay (𝑎) has no effect on effort.
• You can also see this in part a, where the slope of the V(E)
curve just equals the slope of the pay schedule (b) at E*.
(Figure 2.1, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
(Figure 2.3, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford
University Press)
Result 2.1: Agents’ reactions to changes in the employment contract (𝒂, 𝒃) and
productivity (d):
1. For any given contract, more productive agents (with higher 𝑑) will work harder than less
productive agents.
2. Raising the slope parameter (𝑏) of the employment contract will make the agent work harder.
3. Changing the intercept parameter (𝑎) of the employment contract will have no effect on the
agent’s optimal effort level.
Result 2.2: Agents’ reactions to changes in the employment contract (𝒂, 𝒃) with the
𝑬𝟐
baseline production and cost-of-effort functions (𝑽 𝑬 = and d = 1):
𝟐
Before doing the full-blown principal’s problem, we’ll do a simpler, “warm-up” problem first.
As it turns out, this “warm-up” problem is the way most people think about the P-A problem when
they first encounter it.
It gives the wrong answer to the P-A problem. But we’ll learn some useful things from this mistake.
Note: We’ll stick to the baseline production and utility functions for the rest of this chapter.
(We’ll ignore the agent’s participation constraint in this warm-up problem by assuming that the
principal’s most preferred contract turns out to be acceptable to the agent).
Substituting the agent’s response to the contract, 𝐸 = 𝑏, into the definition of profits yields:
max Π = 𝑏 − 𝑎 + 𝑏2 = 𝑏 − 𝑏 2 − 𝑎 = 𝑏 − 𝑏 2
𝑏
Thus, 𝑏 ∗ = 0.5
⟹ a 50% (of net revenues generated by the agent) commission rate maximises profits in this
situation.
More generally:
• More generally (beyond the baseline production and utility functions):
the profit-maximising 𝒃 is always strictly between zero and 1 for any production and
utility function.
Why:
• 𝑏 = 0 yields no profits because the agent does nothing
• 𝑏 = 1 yields no profits (despite high effort) because it gives all the profits away.
• Thus, in this warm-up problem, the profit-maximising commission rate trades off two competing
goals:
• -incentives (higher 𝑏 raises effort)
• -distribution (higher 𝑏 redistributes income from the principal to the agent).
A final thing to notice about our ‘warm-up’ solution to the P-A problem:
(Figures 3.1 and 3.2, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
A final thing to notice about our ‘warm-up’ solution to the P-A problem:
• Principal wouldn’t mind very much (because profits are a relatively flat function of b at 𝑏 = .5)
• Agent would like it a lot, (utility is convexly increasing in 𝑏)
Thus, starting at 𝑏 = .5, a small increase in 𝑏 benefits the worker but doesn’t really hurt the firm.
This suggests that efficiency might be improved by raising 𝒃 beyond . 𝟓….
𝑎𝑙𝑡 𝑎𝑙𝑡 𝑏2
Rearranging and setting 𝑈 = 𝑈 gives 𝑎=𝑈 − (1)
2
STEP 1:
𝑏2
𝑎= 𝑈 𝑎𝑙𝑡 − (1)
2
If the agent always chooses their effort to maximise their utility (taking 𝑏 as given), equation (1) tells us
how much base pay (𝑎) we need to give the agent so they’ll attain the ‘target’ utility of exactly 𝑈 𝑎𝑙𝑡 .
The better off we want the agent to be (i.e. the higher a 𝑈 𝑎𝑙𝑡 we want to achieve), the higher we need to
set 𝑎.
However, since 𝑎 and 𝑏 are alternative ways to make the agent better off, when 𝑏 is higher we don’t need
to give the agent as much 𝑎 to attain the same level of utility.
STEP 2:
max 𝛱
𝑏
Π = 𝐸 − 𝑎 + 𝑏𝐸
= 𝑏 − 𝑎 + 𝑏2 (since 𝐸 = 𝑏)
= 𝑏 − 𝑈 𝑎𝑙𝑡 + 𝑏 2 /2 − 𝑏 2 (since 𝑎 = 𝑈 𝑎𝑙𝑡 − 𝑏 2 /2)
= 𝑏 − 𝑏 2 /2 − 𝑈 𝑎𝑙𝑡
𝜕Π
FOC for a maximum are: = 1 − 𝑏 = 0. (SOC is fulfilled, too)
𝜕𝑏
Therefore, 𝑏 ∗ = 1.
Graphically:
(Figure 3.3, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
Thus, in the profit-maximising contract, the principal should set a commission rate of 100%;
i.e. the agent’s pay should rise by one dollar for every dollar the agent contributes to net revenue.
Notice that this result doesn’t depend on 𝑈 𝑎𝑙𝑡 , i.e. the level at which we choose to set the agent’s
utility.
It follows that a 100% commission rate is profit-maximising, regardless of how well off we want
the agent to be!
To sum up, let’s list the agent’s effort, output, utility, and the firm’s profits, etc. at two different
commission rates (50% and 100%), with 𝑎 set in both cases to guarantee the worker a utility level
(𝑈 𝑎𝑙𝑡 ) of 0.25:
(Table 3.1, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
To sum up:
We can make the principal better off without hurting the agent by switching from the
(a,b) = (1.25, .5) contract to the (a,b) = (-.25, 1) contract.
Lessons:
1. “Put the rewards where the decisions are made”.
When the agent controls a decision that affects the utilities of a larger group (in this case himself
plus the principal), it is profit maximising to have the agent bear all the costs, as well as all
the returns of his actions.
This result does not depend on the level of 𝑈 𝑎𝑙𝑡 —i.e on how well-off we want the agent to be.
2. When agents receive 100% of the fruits of their labour at the margin (𝑏 = 1), principals’ only
source of profits is from setting a negative level of base pay, 𝑎 (in other words by “selling the job
to the worker”).
⟹ Compared to the ‘warm-up’ problem: we can set b =1 to fully incentivise the agent, while a can be
used to achieve any feasible distributional outcome you like.
While this might seem strange at first, there are at least three distinct ways it actually occurs in
today’s economy.
• Taxi drivers, hairdressers at booth rental salons, manicurists, Fedex Ground workers, real estate
agents (desk fee), stock traders at ‘prop’ firms, and…
Many economists refer to the (𝑎 < 0, 𝑏 = 1) solution to the principal-agent problem as the
“franchise solution”.
Today, there are about 800,000 franchisees in the United States: www. statista.com,
Number of franchises has more than doubled in the last 25 years to 48,000 in the United Kingdom:
www.franchise-uk.co.uk
Because that ‘someone’ is their own business, they bear 100% of the costs and rewards associated
with producing their product.
For many workers receiving commission or other forms of performance-based pay, a solution is
to “build the entry fee into the worker’s pay schedule”:
To see how this can work let’s imagine a car salesperson who is paid on the basis of their monthly net
sales, 𝑄.
(From Instructors’ resources, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
But suppose our worker can’t afford to pay up front for the job.
Why not ‘take the entry fee in kind’ by just not paying them for the first n cars they sell each month?
More precisely, let’s replace the previous contract, 𝑌 = 𝑎 + 𝑄 by the contract in bold (see next slide):
(From Instructors’ resources, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
Now suppose our worker is still wary of a contract that doesn’t give them a positive level of base
pay they can rely on each month, regardless of how they perform.
Can we do that?
If 𝑄 < 𝑄0, 𝑌 = 0
If 𝑄 ≥ 𝑄1, 𝑌 = 𝐷 + 𝑏(𝑄 − 𝑄1) where 𝑏 = 1. So the worker still collects their positive ‘draw’, and
earns a 100% commission only on units sold above Q1.
Diagrammatically:
This new schedule also yields exactly the same
output, profits and utility as the original one,
provided that D is not too generous.
(From Instructors’ resources, Personnel Economics, 1e, Peter Kuhn Copyright © 2018 Oxford University Press)
Summing up:
• you get a fixed positive base pay as long as you keep your job
• but you have to sell a minimum amount to qualify for incentive pay
can deliver the exact same results as the superficially ‘extreme’ 𝑎 < 0, 𝑏 = 1 pay plans predicted by
our first principal-agent model.
-to accomplish this, the draw, D can’t be ‘too’ generous relative to output needed to keep your job, Q0.
Economically efficient contracts maximise the sum of the principal’s profits plus the agent’s
utility:
𝑊 = Π + 𝑈.
Alternative terms for 𝑊 include social welfare, social surplus, and the ‘size of the pie’ to be divided
between workers and firms.
𝑊 = 𝑄– 𝑉 𝐸 .
Thus, while payments from the firm to the agent (𝑌) make the firm worse off and the agent better off,
the total amount the firm pays the worker subtracts out of our definition of social
welfare.
That’s because raising or lowering 𝑌 just ‘moves money (or utility) around’ without affecting the total
amount of utility that is produced.
Once those transfers are netted out, social surplus that just equals the total amount
produced (Q), minus the cost of producing it (V(E)).
Problem 1:
Use the same steps as in the ‘warm-up problem’ to find the economically efficient contract under
our baseline assumptions. In other words, choose b to maximise Π + U, subject to the worker’s
incentive-compatibility constraint, to prove the following result:
Problem 2:
Parallel to what we did in the ACTUAL Principal’s Problem, find the contract (a, b) that maximises
the agent’s utility, subject to the incentive-compatibility constraint (𝐸 ∗ = 𝑏) and a participation
constraint for the principal: Π ≥ Π 𝑎𝑙𝑡 .
You’ll find that 𝑏 = 1 again.
Summing up:
The optimal contract between a principal and an agent has a 100% commission rate (b = 1)
regardless of whose welfare (the agent’s, the firm’s, or both) we want to maximise.
This is because it makes sense to maximise the size of the pie regardless of how we
ultimately decide to divide that pie.
Essentially, we maximise the pie by setting b = 1, then divide it up between the parties by picking a
level of 𝑎 that is acceptable to both parties.
For this reason, in most of this module we’ll focus our attention on finding economically efficient
contracts, i.e. contracts that maximise 𝑊 = Π + 𝑈.
• Camerer, Colin F., and Ulrike Malmendier, 2007. "Behavioral Organizational Economics." In
Peter Diamond and Hannu Vartiainen, eds., Behavioral Economics and Its Applications.
Princeton University Press. Pages 235-290.