H1. Why do people stay poor? Part 1
Two Main Views
1. Differences: Ability, talent, motivation, education, etc.
o Policy focus: Equalizing opportunities
2. Poverty Traps: People are poor because they were poor yesterday
o Policy focus: One-off big interventions
Approach to Find the Answer
1. Identify Differences: Compare poor and rich people (health, income).
o Equalize difference between poor and rich; If poor remain poor → Poverty trap
2. Track Over Time: Do richer poor escape poverty?
3. Give Resources: Test if resource transfer helps escape poverty.
The Economical Approach
Step 1: Formalize the Problem
Aim: Simplify theory to allow empirical testing and assumptions.
Focus on total assets: Kit+1=Kit + Iit − dit
Total asset next period= the amount of stuff I had yesterday + the new stuff I bought – the stuff that
broke/ depreciated
o Assumptions:
1. Assets depreciate at constant rate: dit=d⋅Kit
2. Savings = Investment: Iit=s⋅Yit (Yit individual earns during the current period;
focus on a world of self-employment)
o Simplified Model:
Kit+1=Kit+s⋅Yit−d⋅Kit
o Yit = f(kit) depends on:
The more assets assets an individual has, the higher their income is likely to be (simplify the world to
much)
Yit=Ai⋅f(kit) where Ai = productivity
Transition equation; Kit+1= s ⋅ Ai⋅f(kit) + (a-d) ⋅ kit
Variable Direction Effect Reasoning
of Change on
Ki,t+1
s (Savings Rate) Increase ↑ More capital is reinvested into production, boosting future capital.
Decrease ↓ Less reinvestment into production, slowing capital growth.
d (Depreciation Increase ↓ A larger proportion of existing capital is lost each period.
Rate) Decrease ↑ Less capital is lost, allowing more to carry forward.
Ai Increase ↑ Higher productivity increases the efficiency of capital in production.
(Productivity) Decrease ↓ Lower productivity reduces the effectiveness of capital.
Kit (Current Increase Mixed Ki,t: May lead to diminishing returns in f(Kit)(depending on f's functional form)
Capital) but increases the total base for depreciation.
Higher capital increases production, but also increases depreciation due to
(1−d).
Decrease Mixed Lower Kit : Results in lower production but also less depreciation
,Key Questions:
Differences: different Ai will lead to different assets today and tomorrow
Poverty traps: the function f(.) is key
The function f(.)
When someone temporarily injects capital (money), the capital level can increase
in the short term (ST). However, if depreciation remains greater than
investment, the capital will eventually decline back to its original level over time.
This explains why temporary stimuli are not sufficient to lift an economy out of a
poverty trap.
Threshold K∗: If assets grow above this, poverty traps exist.
Existence of some threshold K* implies an S shape
Defend your assumptions
A1 Assets depreciate at a constant rate.
A2 Individuals save a constant proportion of their income and invest all savings.
o are not crucial and do not require extensive justification.
A3 Income is given by Yit = Ai · f (Kit ).
o Ai is constant:
This can be defended by arguing that productivity does not change significantly in the
short term, even though this is more realistic over the long term.
o f(⋅) depends only on Kit:
This can be defended by arguing that other factors, such as labor, are implicitly
accounted for or held constant to isolate the relationship between capital and
income.
Hidden assumption: The absence of fully functioning credit markets is easy to justify, as this
is widely accepted.
Step 2: Test Empirically
Idea: Identify thresholds
o Low levels of Ki0: below the threshold
o High levels of Ki0: above the threshold
-> Where the change happens, is where K* is -> Easy!
Problem: Endogeneity (high assets ≠ random; linked to Ai).
• Endogeneity arises when a variable (e.g., Ki0, the initial asset level) is both influenced by and
influences other factors that also affect the outcome (Kit, the future asset level). People with
initial high levels of Ki 0 are a not a random group of people. On average they will be more
productive i.e. have higher Ai . Higher Ai will also cause higher Kit in the future.
• difficult to establish a causal relationship
Solution: Randomized Control Trials (RCT): Break link between Ai and assets.
o Example: Shock assets randomly: Ki1=Ki0+Ti
Ti = t if treated and Ti = 0 if in the control group
o Compare treated vs. control: Does treated group escape poverty?
, Ethics: RCT may not always be practical or ethical.
• Ethical concerns:
o Participants cannot choose their group, and random allocation may seem unfair.
o Intervening in lives without guaranteed benefits requires careful consideration.
• Unintended harm:
o Treatment groups may misuse assets or face barriers, worsening their situation.
o Control groups may miss out on proven benefits, raising ethical questions.
• Partial aid for control groups: Providing minimal assets to control groups complicates causal
analysis and may lead to resource constraints or perceived inequality.
• Limitations of RCTs: RCTs struggle with long-term, complex, or large-scale interventions.
Findings may not generalize across different contexts or populations.
H2. Why do people stay poor? Part 2
1. Balboni et al. 2022
o Large asset transfer enabled households to switch occupations and escape extreme poverty.
o Findings: Long-lasting returns >>> Initial cost → Strong evidence for poverty trap
explanation.
2. Two Key Questions
o How much are we leaving on the table? (gap between reality and an
ideal situation (first best))
Current economic distribution is not optimal due to frictions:
Transport costs
Incomplete credit markets
Monopolistic power
Labor market frictions
Need to compare current reality to a counterfactual world (e.g., no credit
frictions).
Modeling individual occupation choice: livestock or laborer.
How do choices differ under complete credit markets?
How much on aggregate is the economy better off with the
unconstrained choices?
o Are the results externally valid?
Modeling Misallocation
1. Individuals Choose:
o Hours as laborer (h)
o Hours in livestock (l)
o Hours to hire farm help (h′)
o Constraint: Limited time → Opportunity costs exist.
2. Key Insight
o Total misallocation costs = $15M
o Cost to eliminate poverty traps = $1M
o BUT: General Equilibrium (GE) Effects → Prices might adjust if too many people shift
to livestock.
A catch-all term to describe how prices and other variables may change if large numbers of people change their actions
Two Main Views
1. Differences: Ability, talent, motivation, education, etc.
o Policy focus: Equalizing opportunities
2. Poverty Traps: People are poor because they were poor yesterday
o Policy focus: One-off big interventions
Approach to Find the Answer
1. Identify Differences: Compare poor and rich people (health, income).
o Equalize difference between poor and rich; If poor remain poor → Poverty trap
2. Track Over Time: Do richer poor escape poverty?
3. Give Resources: Test if resource transfer helps escape poverty.
The Economical Approach
Step 1: Formalize the Problem
Aim: Simplify theory to allow empirical testing and assumptions.
Focus on total assets: Kit+1=Kit + Iit − dit
Total asset next period= the amount of stuff I had yesterday + the new stuff I bought – the stuff that
broke/ depreciated
o Assumptions:
1. Assets depreciate at constant rate: dit=d⋅Kit
2. Savings = Investment: Iit=s⋅Yit (Yit individual earns during the current period;
focus on a world of self-employment)
o Simplified Model:
Kit+1=Kit+s⋅Yit−d⋅Kit
o Yit = f(kit) depends on:
The more assets assets an individual has, the higher their income is likely to be (simplify the world to
much)
Yit=Ai⋅f(kit) where Ai = productivity
Transition equation; Kit+1= s ⋅ Ai⋅f(kit) + (a-d) ⋅ kit
Variable Direction Effect Reasoning
of Change on
Ki,t+1
s (Savings Rate) Increase ↑ More capital is reinvested into production, boosting future capital.
Decrease ↓ Less reinvestment into production, slowing capital growth.
d (Depreciation Increase ↓ A larger proportion of existing capital is lost each period.
Rate) Decrease ↑ Less capital is lost, allowing more to carry forward.
Ai Increase ↑ Higher productivity increases the efficiency of capital in production.
(Productivity) Decrease ↓ Lower productivity reduces the effectiveness of capital.
Kit (Current Increase Mixed Ki,t: May lead to diminishing returns in f(Kit)(depending on f's functional form)
Capital) but increases the total base for depreciation.
Higher capital increases production, but also increases depreciation due to
(1−d).
Decrease Mixed Lower Kit : Results in lower production but also less depreciation
,Key Questions:
Differences: different Ai will lead to different assets today and tomorrow
Poverty traps: the function f(.) is key
The function f(.)
When someone temporarily injects capital (money), the capital level can increase
in the short term (ST). However, if depreciation remains greater than
investment, the capital will eventually decline back to its original level over time.
This explains why temporary stimuli are not sufficient to lift an economy out of a
poverty trap.
Threshold K∗: If assets grow above this, poverty traps exist.
Existence of some threshold K* implies an S shape
Defend your assumptions
A1 Assets depreciate at a constant rate.
A2 Individuals save a constant proportion of their income and invest all savings.
o are not crucial and do not require extensive justification.
A3 Income is given by Yit = Ai · f (Kit ).
o Ai is constant:
This can be defended by arguing that productivity does not change significantly in the
short term, even though this is more realistic over the long term.
o f(⋅) depends only on Kit:
This can be defended by arguing that other factors, such as labor, are implicitly
accounted for or held constant to isolate the relationship between capital and
income.
Hidden assumption: The absence of fully functioning credit markets is easy to justify, as this
is widely accepted.
Step 2: Test Empirically
Idea: Identify thresholds
o Low levels of Ki0: below the threshold
o High levels of Ki0: above the threshold
-> Where the change happens, is where K* is -> Easy!
Problem: Endogeneity (high assets ≠ random; linked to Ai).
• Endogeneity arises when a variable (e.g., Ki0, the initial asset level) is both influenced by and
influences other factors that also affect the outcome (Kit, the future asset level). People with
initial high levels of Ki 0 are a not a random group of people. On average they will be more
productive i.e. have higher Ai . Higher Ai will also cause higher Kit in the future.
• difficult to establish a causal relationship
Solution: Randomized Control Trials (RCT): Break link between Ai and assets.
o Example: Shock assets randomly: Ki1=Ki0+Ti
Ti = t if treated and Ti = 0 if in the control group
o Compare treated vs. control: Does treated group escape poverty?
, Ethics: RCT may not always be practical or ethical.
• Ethical concerns:
o Participants cannot choose their group, and random allocation may seem unfair.
o Intervening in lives without guaranteed benefits requires careful consideration.
• Unintended harm:
o Treatment groups may misuse assets or face barriers, worsening their situation.
o Control groups may miss out on proven benefits, raising ethical questions.
• Partial aid for control groups: Providing minimal assets to control groups complicates causal
analysis and may lead to resource constraints or perceived inequality.
• Limitations of RCTs: RCTs struggle with long-term, complex, or large-scale interventions.
Findings may not generalize across different contexts or populations.
H2. Why do people stay poor? Part 2
1. Balboni et al. 2022
o Large asset transfer enabled households to switch occupations and escape extreme poverty.
o Findings: Long-lasting returns >>> Initial cost → Strong evidence for poverty trap
explanation.
2. Two Key Questions
o How much are we leaving on the table? (gap between reality and an
ideal situation (first best))
Current economic distribution is not optimal due to frictions:
Transport costs
Incomplete credit markets
Monopolistic power
Labor market frictions
Need to compare current reality to a counterfactual world (e.g., no credit
frictions).
Modeling individual occupation choice: livestock or laborer.
How do choices differ under complete credit markets?
How much on aggregate is the economy better off with the
unconstrained choices?
o Are the results externally valid?
Modeling Misallocation
1. Individuals Choose:
o Hours as laborer (h)
o Hours in livestock (l)
o Hours to hire farm help (h′)
o Constraint: Limited time → Opportunity costs exist.
2. Key Insight
o Total misallocation costs = $15M
o Cost to eliminate poverty traps = $1M
o BUT: General Equilibrium (GE) Effects → Prices might adjust if too many people shift
to livestock.
A catch-all term to describe how prices and other variables may change if large numbers of people change their actions
