should robots be taxed? discussion by lukas mayr
TRANSCRIPT
Should Robots be Taxed?
Joao Guerreiro, Sergio Rebelo and Pedro Teles
Discussion by Lukas Mayr
November 17, 2017
The Paper
Routine Labor Non-Routine Labor
2
The Paper
Routine Labor Non-Routine LaborRobots
3
Decreasing costs of automation
Routine Labor Non-Routine Labor Robots
4
Tax robots ...
Routine Labor Non-Routine Labor Robots
5
... or let automation happen?
Non-Routine Labor Robots
6
Mirrleesian taxes
PropositionIn the optimal plan, when automation is incomplete (m < 1) robot taxesare strictly positive (τx > 0)
Increase in τx reduces the relative wage wn/wr
Makes non-routine worker less inclined to mimic the routine worker
???
Incentive compatibility constraint for non-routine workers
u(Cn)− v(Nn) ≥ u(Cr)− v(wrNr
wn
)
7
Mirrleesian taxes
PropositionIn the optimal plan, when automation is incomplete (m < 1) robot taxesare strictly positive (τx > 0)
Increase in τx reduces the relative wage wn/wr
Makes non-routine worker less inclined to mimic the routine worker
???
Incentive compatibility constraint for non-routine workers
u(Cn)− v(Nn) ≥ u(Cr)− v(wrNr
wn
)
7
Mirrleesian taxes
PropositionIn the optimal plan, when automation is incomplete (m < 1) robot taxesare strictly positive (τx > 0)
Increase in τx reduces the relative wage wn/wr
Makes non-routine worker less inclined to mimic the routine worker
???
Incentive compatibility constraint for non-routine workers
u(Cn)− v(Nn) ≥ u(Cr)− v(wrNr
wn
)
7
Mirrleesian taxes
PropositionIn the optimal plan, when automation is incomplete (m < 1) robot taxesare strictly positive (τx > 0)
Increase in τx reduces the relative wage wn/wr
Makes non-routine worker less inclined to mimic the routine worker
???
Incentive compatibility constraint for non-routine workers
u(Cn)− v(Nn) ≥ u(Cr)− v(wrNr
wn
)
7
Mirrleesian taxes
PropositionIn the optimal plan, when automation is incomplete (m < 1) robot taxesare strictly positive (τx > 0)
Increase in τx reduces the relative wage wn/wr
Makes non-routine worker less inclined to mimic the routine worker
???
Incentive compatibility constraint for non-routine workers
u(Cn)− v(Nn) ≥ u(Cr)− v(wrNr
wn
)
7
Routine Worker: Assembly Line Worker
8
Non-Routine Worker: Engineer
9
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Example
Fabric worker (assembly-line work) and engineer
With τx = 0: wr = 10 USD/hour and wn = 40 USD/hour
⇒ Engineer gets 40 USD if he works an hour on the assembly line
With τx = 20%: wr = 15 USD/hour and wn = 30 USD/hour
⇒ Engineer gets 30 USD if he works an hour on the assembly line
⇒ Implicit assumption: Engineer always has the same marginalproductivity on the assembly line as he has with engineering
10
Mirrleesian taxes
Fixing the incentive compatibility constraint:
u(Cn)− v(Nn) ≥ u(Cr)− v(κNr)
κ is constant
Most sensible parameterization is κ = 1
⇒ Never tax robots!!!
11
Some additional remarks
Interesting and highly relevant question
Nice model set up with routine tasks as varieties
In Ramsey problems suddenly no informational problem anymore
HSV tax function but no transfers delivers result of no fullautomation and positive optimal robot taxes almost by assumption
o Not sure if this is really the appropriate status quoo There are at least some government transfers
Endogenous human capital formation
o Keeping demand for routine workers high by taxing robotsdiscourages human capital accumulation
12
Some additional remarks
Interesting and highly relevant question
Nice model set up with routine tasks as varieties
In Ramsey problems suddenly no informational problem anymore
HSV tax function but no transfers delivers result of no fullautomation and positive optimal robot taxes almost by assumption
o Not sure if this is really the appropriate status quoo There are at least some government transfers
Endogenous human capital formation
o Keeping demand for routine workers high by taxing robotsdiscourages human capital accumulation
12
Some additional remarks
Interesting and highly relevant question
Nice model set up with routine tasks as varieties
In Ramsey problems suddenly no informational problem anymore
HSV tax function but no transfers delivers result of no fullautomation and positive optimal robot taxes almost by assumption
o Not sure if this is really the appropriate status quoo There are at least some government transfers
Endogenous human capital formation
o Keeping demand for routine workers high by taxing robotsdiscourages human capital accumulation
12
Some additional remarks
Interesting and highly relevant question
Nice model set up with routine tasks as varieties
In Ramsey problems suddenly no informational problem anymore
HSV tax function but no transfers delivers result of no fullautomation and positive optimal robot taxes almost by assumption
o Not sure if this is really the appropriate status quoo There are at least some government transfers
Endogenous human capital formation
o Keeping demand for routine workers high by taxing robotsdiscourages human capital accumulation
12
Some additional remarks
Interesting and highly relevant question
Nice model set up with routine tasks as varieties
In Ramsey problems suddenly no informational problem anymore
HSV tax function but no transfers delivers result of no fullautomation and positive optimal robot taxes almost by assumption
o Not sure if this is really the appropriate status quoo There are at least some government transfers
Endogenous human capital formation
o Keeping demand for routine workers high by taxing robotsdiscourages human capital accumulation
12