emigration & prices
DESCRIPTION
Emigration & Prices. M2R Thesis Economie de la Mondialisation Supervisors: Biagio Speciale / Mathilde Maurel Georgette A. Fernandez Laris. Agenda. Revisiting research question and model specification Database construction Empirical Strategy & First results. General Question: - PowerPoint PPT PresentationTRANSCRIPT
Emigration & Prices
M2R Thesis Economie de la MondialisationSupervisors: Biagio Speciale/ Mathilde Maurel
Georgette A. Fernandez Laris
Agenda
Revisiting research question and model specification
Database construction
Empirical Strategy & First results
General Question:
- Examine the impact of Mexican emigration on goods prices.
Benchmark Paper:
“The Effect of Low Skilled Immigration on U.S. Prices: Evidence from CPI Data”
Cortes, Patricia. 2008. “The Effect of Low Skilled Immigration on U.S. Prices: Evidence from CPI Data” Journal of Political Economy 116 (3): 381-422
Research Question RevisedQuestion What is the impact of Mexican emigration on good’s prices?
Motivation: - Fill-in research gap:
- Vast literature on U.S – Mexico Migration- Destination country focus: impact on the labor market outcomes of local or native U.S
workers- Source country focus: concentrating on emigration’s impact on remittances, labor market
participation and network effects.
- No earlier attempts to examine the impact of emigration on Mexican prices
Relevance: Possibility to test new mechanisms through which emigration impacts the unequal distribution of standards of living across Mexico .Examine new ways in which emigration can affect the real purchasing power of Mexicans who stay.
Labor Market
Participation rates at source
Remittances
EducationEntrepreneurship
Health
Model SpecificationPrice Regression (8 different categories)
= ) + + +
() xj:municipality
i: city agglomerate, composed of many representative municipalities (i=1, 2,..., 45)j: municipality (j=1, 2,…, 475)t: time period (t=1, 2)
- First time period 1995-2000- Second time period 2005-2010
Benchmark (Cortes’)
Database ConstructionI. Prices II. Emigration
Source: Mexican Central Bank (Banco de Mexico) & National Institute of Statistics and Geography (INEGI) http://www.banxico.org.mx; http://www.inegi.org.mx/sistemas/bie
- Dependent Variable ConstructionConsumer Price Indexes for eight product categories classified by object of expenditure
= ) + + +
Specific goodijk-----Ex: Cereal brand X, 400g sold at store Y Generic goodjk-----Ex: Cereal in flakesSubindexk----Ex: Food, beverages and tobacoINPC= CPI
I. PricesII. Emigration
- Instrument Data
Source: Woodruff and Zenteno’s Historic State Migration rates (1955-1960)Fraction of each Mexican state’s population that migrated to the United States over period.
- Main Explanatory Variable: municipal level data on emigration Source: National Council of Population (CONAPO) & National Institute of Statistics and Geography (INEGI)http://www.conapo.gob.mx; http://www.inegi.org.mx/sistemas/bie
=
= ) + + +
Matching
Percentage of households with emigrants to U.S by municipality, 2010
Trends & Stylized Facts
Aggregation Level % Hhds with
remittance receits % Hhds with
emigrant % Hhds with
circular migrants % Hhds with
return migrants
National Average
2000 4.47 3.95 0.97 0.87
2010 3.63 1.94 0.92 2.19
State Average
2000 4.93 4.19 1.08 1
2010 4.06 2.02 1 2.46
Municipality Average
2000 6.65 6.04 1.24 1.19
2010 6.5 3.81 1.2 3.45
Biased OLS estimates likely due to:- Omitted Variables EndogeneityUnobserved economic factors and shocks affect prices and emigration rates simultaneously.
Negative economic unobserved factors related to lower economic conditions across regions might also induce greater emigration rates (negative correlation of error with main explanatory variable)
Negative economic shocks, for example 1995 economic crisis, caused increase in contemporaneous emigration and lagged effects on prices
Both make it hard to disentangle true effect of emigration on prices- Reverse Causality EndogeneityFeedback effects from prices to emigration:
Very high costs of living induce higher emigration in region (perhaps due to larger marginalization of population), leads to spurious positive correlation between emigration and prices.
Main Empirical Strategy Challenge:= ) + + +
Model Specification
Own Construction
In practice:
Cortes’ Instrument
Proportion of immigrants to U.S from source country j settling in city i in 1970
Total immigrants to U.S from source country j at decade t
i: U.S destination cityj: source country of origint: time decade (1980, 1990, 2000)
Interpretation:If 1/3 of Mexican immigrants in 1970 in Los Angeles, instrument allocates 33% of all 1990’s Mexicans immigrants to Los Angeles.
IV Solution: Imputed Emigration Rates - Non-experimental approach based on Corte’s instrument construction & using historic state emigration rates
() x
∑𝑗
h𝐻𝑜𝑢𝑠𝑒 𝑜𝑙𝑑𝑠 𝑗𝑡𝑒𝑚𝑖𝑔
Author provided historic emigration ratesSource: Woodruff & Zenteno’s
𝐼𝑛𝑠5559=[1+(𝑚𝑖𝑔5559 𝑥∑𝑗
h𝐻𝑜𝑢𝑠𝑒 𝑜𝑙𝑑𝑠❑𝑒𝑚𝑖𝑔
𝑗𝑡❑ )]
Price Regressions (5 year average prices)
(OLS 1)1 (IV 1)
2 (OLS 2) (IV 2)
3 (OLS 3) (IV 3)
4 (OLS 4)
(IV 4)
CategoryDescription
FoodBeverages & Tobacco
p_cat1_av
Clothing Shoes & Accessories
p_cat2_av
Housing
p_cat3_av
Domestic Accessories:Furniture, appliances, utensils
p_cat4_av
Dependent Variable
l_hhemig -0.00118 0.0176*** 0.00189 0.0593* 0.0177*** 0.00122 0.0109 0.0706** (0.00167) (0.00644) (0.00802) (0.0327) (0.00334) (0.0162) (0.00692) (0.0309)
Municipality Yes Yes Yes Yes Yes Yes Yes Yes
Year
Yes Yes Yes Yes
Yes Yes Yes Yes
Observations 946 946 946 946 946 946 946 946R-squared 0.998 0.998 0.951 0.946 0.992 0.992 0.966 0.957
All variables in logs Standard errors clustered at the municipality x year level are reported in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Results: Instrument
Price Regressions (5 year average prices)
(OLS 5)5 (IV 5)
6(OLS 6)
(IV 6)
7
(OLS 7)
(IV 7)
8 (OLS 8)
(IV 8)
Category Description
Health &Personal Care
p_cat5_av
Transportation: Public & Private
p_cat6_av
Education &Entertainment
p_cat7_av
Other Services
p_cat8_av
Dependent Variable
l_hhemig 0.00730 0.127*** 0.00803*** 0.0374*** 0.0145*** 0.145*** 0.0136*** 0.0815***
(0.00518) (0.0370) (0.00291) (0.0128) (0.00456) (0.0372) (0.00519) (0.0193)
Municipality Yes Yes Yes Yes Yes Yes Yes Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes
Observations 946 946 946 946 946 946 946 946
R-squared 0.988 0.973 0.996 0.995 0.993 0.979 0.985 0.981
All variables in logs Standard errors clustered at the municipality x year level are reported in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Justifying the Instrument Use of Historic State Emigration rates: - Based on literature’s findings:
Strong historical persistence in regional migration behavior in Mexico
- How does it work?
- Assessing Instrument Relevance (Strength)- First-Stage F-statistic
Early historic patterns
Migrant networks
(Social Capital)
Make migration more feasible
Lower migration costs
More emigration
to U.S1920’s railroad & 1950’s Bracero Program
(guest worker program)
Prob > F = 0.0000 F( 1, 945) = 20.18Angrist-Pischke multivariate F test of excluded instruments: Prob > F = 0.0000 F( 1, 945) = 20.18F test of excluded instruments:
Justifying the Instrument Assessing Exogeneity: Cannot directly test exogeneity because I have 1 endogenous regressor and 1 instrument
Does instrument fulfill exclusion restriction?
Concern JustificationHistoric emigration rates might affect current prices through channels other than more recent emigration rates:
• Long history of emigration at community level might induce different patterns of economic development & wellbeing in community which in turn might affect prices.
- Basket of goods used to construct current CPI’s is considered at national level and it has changed a lot uniformly across regions and through time.
- Effects of long history of emigration on current prices get diluted with other general equilibrium effects for tradables
Development of the railroads in certain states and communities ushered in their subsequent development---- unleashing series of factors affecting economic conditions and commercial activity
We control for this using linear fixed effects
* Regional rainfall shocks, as instruments would be much more likely to affect prices through channels other than emigration itself (even using lagged rainfall rates)
Price Regressions (5 year average prices)
(OLS 1)1 (IV 1)
2 (OLS 2) (IV 2)
3 (OLS 3) (IV 3)
4 (OLS 4)
(IV 4)
CategoryDescription
FoodBeverages & Tobacco
p_cat1_av
Clothing Shoes & Accessories
p_cat2_av
Housing
p_cat3_av
Domestic Accessories:Furniture, appliances, utensils
p_cat4_av
Dependent Variable
l_hhemig -0.00118 0.0176*** 0.00189 0.0593* 0.0177*** 0.00122 0.0109 0.0706** (0.00167) (0.00644) (0.00802) (0.0327) (0.00334) (0.0162) (0.00692) (0.0309)
Municipality Yes Yes Yes Yes Yes Yes Yes Yes
Year
Yes Yes Yes Yes
Yes Yes Yes Yes
Observations 946 946 946 946 946 946 946 946R-squared 0.998 0.998 0.951 0.946 0.992 0.992 0.966 0.957
All variables in logs Standard errors clustered at the municipality x year level are reported in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Results: Price effects
Price Regressions (5 year average prices)
(OLS 5)5 (IV 5)
6(OLS 6)
(IV 6)
7
(OLS 7)
(IV 7)
8 (OLS 8)
(IV 8)
Category Description
Health &Personal Care
p_cat5_av
Transportation: Public & Private
p_cat6_av
Education &Entertainment
p_cat7_av
Other Services
p_cat8_av
Dependent Variable
l_hhemig 0.00730 0.127*** 0.00803*** 0.0374*** 0.0145*** 0.145*** 0.0136*** 0.0815***
(0.00518) (0.0370) (0.00291) (0.0128) (0.00456) (0.0372) (0.00519) (0.0193)
Municipality Yes Yes Yes Yes Yes Yes Yes Yes
Year Yes Yes Yes Yes Yes Yes Yes Yes
Observations 946 946 946 946 946 946 946 946
R-squared 0.988 0.973 0.996 0.995 0.993 0.979 0.985 0.981
All variables in logs Standard errors clustered at the municipality x year level are reported in parentheses
*** p<0.01, ** p<0.05, * p<0.1
What lies ahead…
- Possible Robustness checks- Examining division of Tradables vs Non-tradables- Examining channels
What are the main channels through which the price effect mechanism runs?
Q & AThank You