Readings LT Flashcards
(Does Management Matter. Bloom et al)
Special setting and data specifics?
It’s set in India looking at large, multi-plant textile firms.
Treatment plants receive 5 months extensive management consulting from large int. consulting group.
Data compiled from direct observation at the plant/factory
For non-project firms, data was collected during the interview from direct factory observation and discussion with the managers.
(Does Management Matter. Bloom et al)
Empirical design?
66 potential subject firms with 34 expressing interest in project and 17 committing to consulting program.
11 treatment firms with 14 treatment plants and 6 control firms with 1 control plant
Consulting treatment had 3 phases (diagnostic, implementation and measurement)
Estimate the impact of consulting services on management practices via ITT equation. ***** GO OVER
(Does Management Matter. Bloom et al)
Key findings?
Improvement of management practices in treatment plants. Control plants also improve slightly after diagnostic phase and even non-experimental plants in treatment benefit (evidence of cross-plant learning).
Improvements in management led to ++ productivity (from ++ quality and ++ efficiency and reduced inventory) Also followed by longer-run increase in firm size - no plants per firm is higher for better managed firms.
No sig change in invetory during implementation phase - it was only post treatment it was reduced.
Same for output where it increased after.
Profits per plant +++
Major initial barrier was information about whether the practices would be profitable which is why they didn’t take them up initially.
(Does Management Matter. Bloom et al)
Conclusions drawn?
Intervention v expensive for free consulting.
Not strong external validity (other industries, countries)
Firms weren’t implementing changes due to lack of info so maybe should focus on training programs for basic operations management to improve firms productivity
Policies could maybe deal w what may be source of poor management practices (reduce barriers to competition, improve legal/regulatory system to reduce reliance on social capital for trust in management)
(Retail globalization and Household welfare. Atkin et al)
Special setting/Data?
Mexico where no. foreign supermarkets close to quadrupled from 2002 to 2014.
Data on store locations and dates of opening from Mexico’s national association of retail business.
Monthly micro data (price quotes) to create Mexican CPI
The consumer panel micro data of large international market research company.
Also uses Mexican Nat Income & expenditure surveys to view income and sources of income for each houses + expenditure.
(Retail globalization and Household welfare. Atkin et al)
Empirical design?c
Expression: Change in household welfare due to retail FDI on household welfare in the municipality of entry. Decompose this into 6 main channels to find the total welfare effect:
- effect on consumer prices at pre-existing domestic retailers
- effect due to exit of domestic retailers
- all consumer gains derived from being able to shop at foreign store itself
- retail labour income effect
- retail profits effect for domestic store owners
- indirect effect on other sources of household income from nonretail sectors of local economy
Then use household-level income shares by occupation/sector and consumption shares across products/stores to estimate remaining key parameters.
Combine the estimated effects and estimated parameters and the expression to quantify welfare effects of foreign entry for every household and obtain full distribution of welfare effects across households.
Change in welfare due to entry of foreign supermarket expressed as compensating variation (CV) for each household h in municipality of entry.
CV - change in exogenous income necessary for a HH to maintain same utility when a foreign retailer arrives between period 0 and 1
(Retail globalization and Household welfare. Atkin et al)
Key findings?
Prices:
- foreign stores charge 12% lower prices for identical bar codes compared to domestic retailers
Quality:
- foreign stores offer product mix significantly higher quality
Expenditure:
- foreign stores capture more than 30% total household retail expenditure after entering
Distribution effects:
- richest income groups substitute much more of their retail consumption in foreign stores than poor
- large welfare gains for average household which is driven by reduction of cost of living
Effect on domestic retailers:
- negative significant effect on no. of them and also on their profits
(Retail globalization and Household welfare. Atkin et al)
Conclusion?
Retail v important for employment, production and consumption in developing countries
Development should focus more on the services sector
(The Effects of Rural Electrification on Employment. Dinkelman)
Special setting?
South Africa where grid infrastructure was rolled-out and was rapid and extended in rural areas targeting low capacity households instead of industrial users.
(The Effects of Rural Electrification on Employment. Dinkelman)
Data?
Panel dataset of community aggregate variables using 1966 and 2001 South African census data and also: spatial data on local of electrification infrastructure, administrative data on project placement and also measures of geography at baseline.
Unit of analysis for IV strategy is community-year.
(The Effects of Rural Electrification on Employment. Dinkelman)
Empirical design?
- Instrumenting for project placement with land gradient.
- identification strategy: higher gradient raises ave cost of household connection so gradient is an important factor for prioritising areas for electrification but it doesn’t effect employment growth independently - Fixed effects strategy
- to estimate labour market effects using only within-district variation: construct 4-period panel of magisterial districts from cross-sectional household survey data and address non-random project placement and confounding economic trends by directly controlling for magisterial district fixed effects and trends
(The Effects of Rural Electrification on Employment. Dinkelman)
Key findings?
First stage:
- as gradient increases the probability of receiving Eskom Project falls across columns and size of this coefficient doesn’t change much with more controls whilst the precision improves
Employment:
- Female employment increases with electricity project: positive coeffs on poverty rate, sex ratio, female-headed households shows fem employment rises faster in poorer places.
- Male employment rises but not significant.
- frees up women’s time for the market
- could be driven by migration behaviour to electrified places
Electric lighting:
- rises in community with electricity project
Home production:
- cooking with electricity rises
- areas chosen to be electrified because of flatter gradient use electric lighting much more and cooking with wood falls a lot here
(The Effects of Rural Electrification on Employment. Dinkelman)
Conclusion?
- limited external validity - these effects should be interpreted in existing economic conditions of South Africa at the end of Apartheid
- such infrastructure projects likely to increase quality of life and contribute to economic development
(Political Economy of Deforestation in the Tropics. Burgess et al)
Special settings?
Indonesia where it has the largest strands of tropical forest and experiences rapid deforestation.
Also is the 3rd largest producer of greenhouse gases worldwide.
Indonesia also experienced remarkable increase in the no. of administrative divisions over the past decade after the collapse of New Order regime.
(Political Economy of Deforestation in the Tropics. Burgess et al)
Data?
- so much deforestation is result of illegal logging so dataset is constructed from satellite imagery (430 image inputs for each 250by250 pixel used to estimate forest cover loss per year for that pixel)
- dataset combined with data on district boundaries and land-use classifications
- use official data for prices because it will be formed by both legal and illegal supply
(Political Economy of Deforestation in the Tropics. Burgess et al)
Empirical design?
- estimate fixed-effects: Poisson quasi-maximum likelihood count model
E(deforest) = province fixed effect + exp (bnumdistrictsinprov + islandxyear fixed effect)
Coefficient b represents semi-elasticity of deforestation with respect to number of districts in the province
Why use Poisson QML than log dependent variable w OLS?
- there are observations where dependent variable = 0 so a count model is more appropriate
OLS regression
- run price/quantity on numdistrictsin prov and fixed effects of wood type by province and by island by year
Examining SR impact of oil and gas rents on illegal logging?
- Same poisson equation but this time PerCapOilandGas inside the exponential
(Political Economy of Deforestation in the Tropics. Burgess et al)
Key findings?
Deforestation rate:
- increases if an additional district is formed within the province
Contemporaneous effect:
- adding one additional district in a province decreases prices and increases quantity of logs felled though the impact on prices is not stat sig
- after 3 years, impact on prices is sig
- estimate demand elasticity of -2.27
Logging and oil and gas rents:
- each $ of oil and gas rents received reduces logging by 0.3% in ST
- in medium run, oil and gas rents and rents from logging are no longer substitutes
- rents from illegal logging and rents from oil and gas revenue sharing are substitutes only in SR and in medium run over half this effect disappears
(Political Economy of Deforestation in the Tropics. Burgess et al)
Conclusions?
Increasing no. of political jurisdictions is not merely driven by changes in allocation of legal cutting rights but is also due to something happening with regard to illegal logging as well
TO counteract corruption:
- strengthen top-down monitoring and enforcement eg increase prob of detection of illegal activity
- provide district gov w alternative sources of rents
(Weather, Climate Change and Death in India. Burgess et al)
Aim?
Testing whether population at different stages of development are affected differently by same weather variation.
(Weather, Climate Change and Death in India. Burgess et al)
Special setting?
Rural vs urban regions in India. Former is predominantly agricultural employment and latter is non-agric
India primarily a rural country (72% in 2000)
(The Digital Provide. Jensen)
Special setting?
Set in Kerala where fishing is an important industry and mobile phone service was gradually introduced from 1997 and by 2001. Over half were using mobile phones to coordinate sales.
(The Digital Provide. Jensen)
Theory?
ICTs may help poorly functioning markets work better and thereby increase incomes and/or lower consumer prices.
Two principles underpin functioning of the market economy:
- 1st fundamental theorem of welfare economics (i.e., competitive equilibria are Pareto efficient)
- The Law of One Price (i.e., the price of a good should not differ between any two markets by more than the transport costs between them)
(The Digital Provide. Jensen)
Identification strategy?
Paper exploits region-by-region rollout of mobile phone coverage in the Indian state of Kerala. No pre-existing differential trend in market outcomes across regions and no other factor could have influenced outcome changes differently across regions
(The Digital Provide. Jensen)
Empirical method?
Two towns with identical number of fishermen
- Fisherman’s catch (x) is a random variable with an identical distribution across individuals and depends on the density of the fish (d)
- Each zone can either be in a high- or low-density state
- On observing their own catch, each fisherman updates their assessment of the state of their catchment zone: a higher catch induces a switch to a non-local market despite paying the transportation costs (this is because of higher expected gain in profits for an expected price difference)
- In a high density state, greater supply will reduce price dispersion in a given market significantly.
There’s also a search technology: where for a cost, fisherman can learn about the catch in both zones and avoid unprofitable switching. This is purchased up to the point where expected gain from arbitrage equals cost of search
Waste arises when the max quantity demanded is less than the total catch
Measuring welfare:
Producer welfare is change in profits (fixed costs don’t change + inelastic supply)
Consumer welfare is change in consumer surplus from an estimate of the consumers’ demand curve (pre- and post-mobile) and consumer retail
(The Digital Provide. Jensen)
Data?
Weekly surveys of fishermen in Kerala’s 3 northern districts from 1977-2001
Rapid spread of mobile phones with penetration rate reaching 60-75% in around 20 weeks
Broken into 4 time periods – 0 (no mobiles), 1 (mobile phones in region 1), 2 (phones in regions 1 and 2), 3 (phones in all regions)K
(The Digital Provide. Jensen)
Key findings?
30-40% of fishermen on average start selling outside their local market despite still fishing in their zone
Addition of phones results in decline in price dispersion.
Incidence of waste declines to almost 0
Before mobile phones the LOP was violated in almost half cases but it has been reduced dramatically to almost 3-5%
Mobiles increased quantity sold resulting from the decline in waste.
Average profits also increased and there was increase in profits for both fishermen with and without phones – positive externality effect
NET GAIN IN WELFARE FOUND
(The Digital Provide. Jensen)
Conclusions?
Information makes markets work and markets improve welfare
Such technology can increase earnings, which in turn, leads to improvements in health and education
Private sector initiatives: sustainable in the long-run as both parties benefit (suppliers and customers unlike government or NGO projects)
(Railroads of the Raj. Donaldson)
Special setting?
Building of vast network of railroads in colonial India from 1853-1930 by the British government. Investigate how large these benefits of infrastructure was.
(Railroads of the Raj. Donaldson)
Theory?
GE Ricardian trade model with many regions and many commodities where trade occurs at a cost
- Regions have incentives to trade in order to exploit comparative advantage
- Producers can sell more of what they are best at producing
- Model delivers 4 key predictions which are taken to dataset on railway expansion and economic activity in Indian districts
(Railroads of the Raj. Donaldson)
Empirical analysis?
4 step empirical analysis based on the 4 key predictions of Ricardian model:
- Inter-district price differences are equal to trade costs: this allows author to estimate trade costs and to test whether railways reduced the cost of trading in India
- Fixed bilateral trade costs should reduce bilateral trade flows: author test whether railroad-driven reductions in trade costs increase bilateral trade flows
- When a district is connected to the railway network its real income increases: author finds that railroad access raises real income
- Impact of railroad on welfare in a district is captured by its impact on one endogenous variable (the share of that district’s expenditure that it sources from itself): they regress real income on this sufficient statistic (calculated using parameters from empirical steps 1&2) alongside regressors from empirical step 3
(Railroads of the Raj. Donaldson)
Key findings?
Increase in effective distance increases salt price at the destination.
The lowest-cost route measure is estimated to reduce bilateral trade with an elasticity of -1.6
In the average district the arrival of railroad is associated with a rise in income. No bias due to endogenous railroad placement. Railroad dummy is statically significant and implies that a little over ½ of the total impact of railroads estimated can be explained by enhanced opportunities for comparative advantage.
(Railroads of the Raj. Donaldson)
Conclusions?
Railroads reduced the cost of trading, reduced inter-regional price gaps, and increased trade volumes
We need to research: how transportation infrastructure projects can smooth away the effects of local weather extremes on local wellbeing
(Regulation of Entry. Djankov)
Special setting and theory?
Data set on 85 different countries.
They don’t find that stricter entry is associated with higher quality products, better pollution records or health outcomes or keener competition
Public choice theory vs public interest theory (regulation is good).
Within public choice there is the tollbooth theory (regulation serves politicians/bureaucrats) and the capture view (it serves the industry)
(Regulation of Entry. Djankov)
Empirical design?
Did regressions OLS using different dependent variables (quality standards, water pollution, deaths from accidental poisoning, size of unofficial economy etc)
The independent variables are the log number of procedures and log per capita GDP in dollars.
For evidence of tollbooth theory:
- uses corruption as dependent variable
- independent variables are procedures, time, cost and per capita GDP
For checking effect politically:
- dependent variables using various political indicators and log GDP
- dependent variable is procedures
- from this we can see constrained political measures are exogenous to entry regulation
(Regulation of Entry. Djankov)
Results?
Countries with more autocratic government are more likely to have lots of procedures and regulation. even holding income constant
(Regulation of Entry. Djankov)
Evaluation of paper?
Problems with table VII (one about political factors being linked with entry regulation) - multicollinearity and direction of causation
Multicollinearity of GDP with political variables.
In public choice theory, burdensome regulation reflects transfers from entrepreneurs or consumers, which are likely to be distortionary and hence associated with lower income levels.
Countries may be poor because regulation is hostile to new business formation.
(Can labour regulation hinder economic performance? Besley)
Setting/theory?
Industrial relations in Indian states affecting pattern of manufacturing growth.
Labour regulation affects economic performance through relative price effect (raising MC of employee) and an expropriation effect (discouraging investment)
(Can labour regulation hinder economic performance? Besley)
Empirically:
time series and cross sectional data
overcome OVB from cross sectional variation - within country panel data to control for state and time fixed affects
Regression:
y = a + b + ur(st-1) + x+ e
Where y is outcome variable in state s at time t. r is regulatory measure (which we lag one period to capture gap between implementation) and x are other exogenous variables. a is state fixed effect and b is a year fixed effect.
(Can labour regulation hinder economic performance? Besley)
What did they do to help identification?
- panel data regression
reverse causality solutions:
- matching states by level of unionization - creating pairs for treatment and control states with similar levels of unionization
- IV: 2 instruments - level of unionization interacted with post1977 Dummy and land revenue collection regimes interacted with post-1977
Exploited 1975-1977: Gandhi’s declaration of state of emergency which led to decrease of power for congress party and changes in political power in many states
(Can labour regulation hinder economic performance? Besley)
Results?
- labour regulation strongly positively correlated with workdays lost to strikes and lockouts per worker
- non-agricultural output negatively correlated with labour regulation (specifically registered manufacturing
- regulation no significant effect on workers’ payments
- correlation between pro-worker regulation and urban poverty
(Value of Democracy. Burgess)
Empirical design?
Kenyan districts, ethnic segregation.
Counterfactual road network based on goal of maximizing market potential.
Step 1: graphical inspection of districts of president and Kikuyu vs Kalenjin vs others.
Main outcome variable is share of road expenditure received by district divided by population share. (value 1 means road spending proportional to population)
Second method is regression:
road = y + a + b(coethnic district) + g(coethnic district) x democracy + tX + u
coethnic district takes value 1 for those districts where at least 50% of pop has same ethnic as president
y = district fixed effects a = year fixed effected X = vector of baseline demographic, economic and geographic variables
(Value of Democracy. Burgess)
Findings?
Kenyan districts with ethnicity of president receive 2x expenditure on roads and 5 times length of paved roads built relative to what would be population share.
Biases only prevalent in periods of autocracy.
(Value of Democracy. Burgess)
Analysis?
OVB - shifts not exogenous (other factors correlated with regime changes that have an impact on road building)
Kenya representative of other sub-saharan countries so maybe some external validity
(political economy of government responsiveness. Besley)
Empirical design?
India. 2 period model.
Panel data regression:
g = a + b + ds + y(s)(z) + lz + u
where a and b capture state and year fixed effects
s = gov responsiveness g= government responsiveness z= economic, political, media variables m= media and political variables
(political economy of government responsiveness. Besley)
Results?
Economic factors limited influence on gov responsiveness but positive correlation between newspaper circulation levels and 2 measures of government responses.
Local press published in regional language also forces government to respond.
Greater electoral turnout associated with greater responsiveness. Greater political competition also greater responsiveness and also closeness to election.
