Lesson 2/3: Poverty Traps Flashcards
Explain the two views of poverty and how they relate to poverty traps.
Conditional Convergence View:
unequal acces to opportunity, different fundamentals (SOLOW VIEW)
* People have different fundamentals which determine their occupational choices and earnings
* in the long-run they converge to a steady state determined by the fundamentals
Poverty trap view
Unequal acces to opportunity, same fundamentals
* People have different access to opportunity which determines their occupational choices and earnings
* people with the same fundamentals may converge to different steady stated depending on their initital endowments
What are the policy implications of the Solow versus the poverty trap view.
**Solow View: **
* anti-poverty policies support consumption
* drip-feeding trasnfers will help people climb the hill
Poverty-trap view:
* anti-poverty policies support production
* breaking out of the trap requires a large increase in productive assets
Explain the poverty trap model (set up) as discussed in class (Bandiera et al. 2017)
- each person is born with: one unit of time, wealth endowment E & talent for occupation
- Occupations: 1) causal wage labour (pays w), 2) livestock rearing (requires capital K and yields A f(K)
- Which world we are in deoends on: shape of the production function f(K) and whethet credit markets are perfect (or not)
**Set-Up: **
* capital market imperfections
* non-convexities in the production technology
Production function
* either y= Af(K) or y = w, where 0 < w < Af(K)
-> this makes it possible to get multiple equlibria
-> in the absence of perfect capital markets (otherwise you simply borrow), there can arise a poverty trap
No capital markets: discontinuity at k(bar) and two steady-state equilibria
Way out: if s or w(bar) is high enough, you can save your way our (note: K(t+1) = s(w(bar) + k(t))
Explain why the shape of the production function is crucial to explain and assess poverty traps.
- people start from different levels of wealth and access to productive assets: human capital, physical capital
- you get a hill if returns are higher when assets are low: each small investment allows you to climb the hill (all people with same talent will reach teh same point regardless of where they started) -> concave function
- you get a mountain if you need a large investment to get high returns; small investments do not allow you to climn: identical people will reach different points depending on where they started -> not globally concave function (convex segments)
What are the implications of wealth inequality for the persistence of poverty?
- households expereince different long-run trajectories depending on their initital circumstances
- the long-run persistence of poverty is determined by the initital fraction of households below the threshold
- this is a cross-generational problem: one generations inequality of opportunity determines the next generation’s inequality of outcome: no convergence
-> poverty traps are both unfair and inefficient
What is the empirical challenge in identifying poverty traps?
- the two views lead to observationally equivalent outcomes (if you look at the cross-section of individuals at a certain point in time)
- Poverty trap view: thresholds at which the transition equation changes discontinuosly is an unstable equilibrium and individuals quickly diverge away from it
- hard to disentangle responses in terms of accumulation/decumulation of capital vs time-varying factors when tracking individuals over time
Give some of the key take-aways about poverty traps.
- The shape of the production function - and therefore of the transition equation - is crucial for generating poverty traps
- No single friction is sufficient to trap individuals in poverty
- One combination of factors that can lead to a poverty trap in theory is lumpy investments + imperfect credit markets
- setup cost (handwritten note)
- From here onwards, think of general case as transition equation being “S-shaped” in poverty trap world
- Conditional convergence in Solow world yields observationally equivalent steady states as poverty trap
Draw the mechanisms of asset transfers in the Solow versus the poverty trap world.
For solution refer to slide 51 - 54 in my slidedeck.
What is the research questions of Bandiera et al (2017) and how do they answer it?
Research question:Can enabling the poorest people to take on the activities of their richer counterparts set them on a sustainable trajectory out of poverty?
**Set-Up: **
* Northern Bangladesh: irregular demand for causal wage labour, higher grain prices, extreme poverty, food insecurity
* Occupational structure at baseline: hierarchy of occupation by wealth (poor: agriculture and domestic services, rich: self-employed in lifestock and land cultivation), better jobs require productive assets (set apart rich and poor: 94 times higher)
* Evaluate a nationwide program that transfers livestock assets and skills to ‘ultra-poor’ women
* Large-scale randomized control trial (21,000 households in 1,309 villages) in Northern Bangladesh
* Survey the sample (Treatment + Control groups) four times over a seven-year period
The paper by Bandiera et al. studies a program by BRAC. What is the program and how was it set up? What was its impact?
BRAC’s Targeting the Ultra-Poor program
* Randomly allocated across areas (clusters of villages) -> 40 BRAC branches (1,309 villages)
* Beneficiaries are the poorest women in these villages
* Program transfers a large asset (a cow) and training
* Value of the asset = 1 year of PCE (5× typical microloan) (per capita expenditure)
Program Impacts on Labor Supply and Earnings
* Transformative impacts on the labor activity choices of ultra-poor women. Four years after the transfer:
► 217% more hours to livestock rearing
► 17% fewer hours to agricultural labor
► 26% fewer hours to maid services
➡ Suggests poor women had little work capacity at baseline
- Beneficiaries also accumulate more assets: livestock, land, and business assets (e.g., livestock sheds, rickshaws, vans)
- Earnings increase by 21%, per capita consumption expenditure is 11% higher, and the value of household durables is 57% higher four years post-treatment
➡Large treatment effects
What is the motivation behind the follow-up paper to Bandiera et al. (2017)?
If the poorest households face barrier to entering high-return work activities and this is what keeps them in poverty, we expect program beneficiaries to change their labor allocation and escape poverty once these are removed.
* the estimated average effect supports this hypothesis
* but for some it is not good and they fall back into poverty -> What determines this?
What does the paper by Balboni et al. (2022) discuss?
- Why do people stay poor?
- the paper assesses the distribution of assets (and transition equations) of the program beneficiaries (as discussed in Bandiera et al.) -> essentially the paper tests for poverty traps
- Initial distribution: bimodel asset distribution
- program moves the poorest into the lowest density area
How do Balboni et al. (2022) test for the existence of poverty traps?
- poverty traps and differential productivity produce different transition equations
- a necessaty condition for poverty traps is that the transition equation is not concave
- exploit small differences in baseline assets to estiamte transition equation
- panel data covering 11 years post-transfer and large sample allows them to do this
What do Balboni et al. (2022) find on the existence of poverty traps?
- transition equation is S-shaped
- unstable steady state is at the point of the lowest density
- assets move in opposite direction depending on initial distance to threshold: households who passed the threshold (2/3) were able to accumulate 14% more assets, remaining 1/3 lost 16% of asset value o.a.
What are the identification concerns in Balboni et al. (2022)? And how have they been answered?
Identification is based on differences in initial assets that are extremely small relative to the transfer but not randomized.
1) Correlation between baseline assets k0 and asset shocks delta k -> use control groups to account for shocks
2) Correlation between k0 and program response -> exploit additional variation to compare households with same k0: k(t+1) = sAf(k(t)) + (1-d) k(t) where s is savings rate and A is a productivity parameter
higher s -> lower threshold (higher savings rate)
higher A -> lower threshold
Answer
1) control group has only some variation amongst ultra poor but an overal low bimodal distribution
2) Conditional transition equation is still s-shaped
-> testing for povery traps without using (endogenous?) variation in K0 by splitting the sample either according to s or A (using proxies)
-> still produces similar transition equations
What are the long-run dynamics in productive assets and consumption in Balboni et al. (2022)?
- differences in productive assets grow over time
- changes in composition of assets
- average hap in consumption increases
-> conclude that small differences in initital assets can result in large, permanent differences in living standards
What are Balboni et al. (2022)’s key conclusions?
- poor people are not unable to take on more productive employment activities, they just lack the required capital
- poverty threshol implies that only transfers large enough to push beneficiaries past the threshold will reduce poverty in the long run
- to tackle persistent poverty, need big push policies that tap into people’s talents rather than just propping up their consumption
What possibly outstanding questions are left after Balboni et al. (2022)?
- Do these results replicate in other settings?
- Can the **graduation model **be tweaked and produce similarly transformative impacts?
- How does it compare with other poverty alleviation programs?
Explain the concept of universal basic income as a poverty alleviation tool.
- Banerjee et al. (2023) ran the first randomized evaluation of universal basic income (UBI) in a developing country, covering every adult in targeted villages in Kenya for up to 12 years
- Aim: asses impacts on labor supply, earnings, and well-being, also comparing the program to alternative transfer models like short-term UBi and lump-sum payments
**Findings: **
* increase in entrepreneurship -> switch away from wage work
* incentive to work not reduced, maybe increased
* big increase in number of enterprises
* consumption increase
* increase in incomes
* increase in wages
* reduction of depression
* no increase in inflation
* more product versatality
* land prices go up
–> slope of the transition equation changed (before the discontinuity)
