Topic 3: Political Economy Flashcards
Alesina and Fuchs-Schundeln (2007)
Effect of communism on people’s preferences for social policies => whether exposure to communism 50 years ago has an effect today
Dependent variable: preferences for an intrusive state in providing services and redistribution
Crucial assumption:
1. omitted relevant variables could confound the level of y - unobservables should not explain our results
2. simultaneity should not bias an estimator e.g. peoples preferences should not affect the regime - e.g people choose institutions and cannot say about whether people like to live in a communist state when they chose to live in it
=> how to rule these out? exogenous shock!
Alesina and Fuchs-Schundeln use an exogenous shock of the splitting of Germany after WWII and then reunification in 1990.
Treatment Group: East Germans (exposed to Communism)
Control Group: West Germans
=> study whether exposure to communism had an effect on people’s preferences.
Main results:
1. Exposure to 45 years of communism instilled in people the view that the state is essential in intervening
=>This contrasts the alternative hypothesis that people who were really oppressed would be pro-free market after the fall of communism
- Empirical concerns: it must be that pro-state preferences were developed during the exposure and not pre-determined => it must be that pre-WWII there were no differences in characteristics between east and West
=> solution: voting behaviour pre WWII=> found no significant differences - Still differential effect, although over time the East Germans became more pro-market, since the interaction between East and 2002 was negative and statistically significant => would take 1 to 2 generations for their views to align
- also would expect that older people who were more exposed to communism would have stronger pro-state preferences => confirmed and statistically significant
Narciso and Severgini (2017)
Whether Irish Famine induced the development of anti-British sentiment
Famine: huge shock => Irish were heavily specialised in potatoes and potato blight reduced production and therefore people starved. The British government was not providing enough relief => Irish Revolution 1916
Advantage of paper: detailed information and merged two sources of data for the Irish Census and the Military Archives
- The famine led to higher emigration from the countries most affected and so the authors control for emigration
Results:
- Even controlling for countries with emigration, highly intense famine increased the probability for rebellion
- Focal effect: a positive and statistically significant effect of the famine for rebellious action, especially if Catholic and can speak Gaelic - more probable to become rebellious by identifying with the Irish people and developing anti-British sentiment.
- Being born in an area where the famine was more intense increases the probability of becoming a rebel
Endogeneity issues:
- Problem of misrecording of data– contain measurement error or missing information => omitted relevant variables:
2 stages:
- IV estimation to use exogenous variation of the famine - 15 instruments for proxies on the potato blight, including the weather, precipitation and seasons (e.g. summer, spring, winter)
- IV Probit estimation
=> bias is not an issue, as results are similar to the baseline
=>the great famine in the 1840s was a great historical shock that negatively impacted the view of the Irish towards the British
Becker and Pascali (forthcoming)
Can the presence or absence of complementarities in the labour market between the Jewish minority and the majority populations explain the variation in the anti-Semitic sentiments over time and regions?
Can frictions in the labour market explain inter-ethnic tensions and anti-Semitism?
Historical background:
1. Catholic ban of money lending but the Jews were allowed to do so => Jews specialised in the credit market because the ban did not apply to them
2. Protestant reformation => split Germans between Catholics and Protestants. Those who were Protestants, could trade in the credit market => friction
=> Whereas in Catholic areas, complementarities between the Jewish minority persisted, in Protestant areas, Jews lost their prerogatives in credit sector
Hypothesis:
Anti-Semitism increased in Protestant areas relative to Catholic areas because of the friction of the credit market after the protestant reformation
=> Compare the change between protestant and catholic in the post-treatment relative to the pre-treatment
2 contributions of this paper:
- Estimation strategy => diff-in-diff, panel dataset on pogroms and other anti-Semitic behaviour observed from every century from 1300 to 1900
- Data => impressive wealth of data on anti-semitic books, support to anti-Semitic parties, etc.
Empirical Strategy 1:
1. Diff-in-diff with city-level panel data testing the differential effect of anti-Semitic acts between Protestant and Catholic cities in the period after the reformation relative to before
=> Effect: positive and statistically significant=> friction in the labour market induced by the Reformation increased anti-Semitic sentiment
Also checked for whether there is a pre-trend before the Reformation => not statistically significant, no pre-trend => genuine shift after the reformation => structural change
***Robustness checks:
Confounding factors e.g. war activity- some cities might be more violent than others or more exposed to war which increased the prob. of anti-Semitic sentiment and education and population
Controls for war activity and educ=> still statistically sig => the main estimator remains unaffected
Empirical Strategy 2:
Number of books for anti-Semitic content as dependent variable => measure the demand for anti-Semitic content and therefore sentiment
=> Following the Protestant Reformation => number of anti-Jewish books published increased by 7 percentage points
**Finally…
City development might be an unobservable factor: maybe the Protestant Reformation induced differences in Anti-Semitism dependent on the development of the city and Jewish lending before 1500 might measure the success of the city’s economy.
After controlling for population => becomes insignificant but maybe because adding more centuries increases the variation => variance=> reduces the significance of β.
=> highly significant and persistent effect of the Protestant reformation on the Anti-Semitic sentiment relative to the pre-treatment period and increase in ethnic hostility
Belmonte and Di Lillo (2018)
Inter-Ethnic tensions in South-Tyrol
Exogenous factor: the Italianization of South Tyrol during fascist Italy
Historical background: persistence of cultural assimilation => tried to homogenise clues, ban of other languages other than the official one => ethnic minorities disappear
The Italianisation of South Tyrol, which produced persistent distortions in favour of Italians (comparative advantage). in 1966 - time discontinuity, where there was a policy reformation - a government more sensitive to the German minority redistributed jobs result in increased tension between Germans and Italians
Hypothesis:
Whether after the announcement of the reform (1966) in municipalities where Italians were fewer, increased anti-German sentiment
In municipalities where Italians made up 20 percent of population, they were highly specialised and more than 80 percent were public servants => Italians replaced Germans extremely specialised in the public sector
1966: time discontinuity - policy reformation - change in governments more sensitive to ethnic minorities - redistribution proportional to numerosity of each group - side by side with Italians => friction in the labour market on inter-ethnic tensions??
Excessive data on political attitudes, votes for fascist parties.
South Tyrolean setting particularly helpful - two main parties - DC and PCI. MSI was founded in 1946 as a post-fascist party- defence of the italianness of South Tyrol).
Diff-in-Diff strategy: focusing on the share of Italians and the post-reform period relative to pre.
Results:
one standard deviation below the mean has a positive and significant increase in the vote share of the MSI
Endogeneity issues: likely to be biased if people chose where to settle down and maybe less impact in villages with Italian communities less sensitive to the inter-ethnic conflict
- Collected data from migration flows (stations along the railway, share of houses built by the fascist regime, etc.)
- Strong, persistent relationship between the Italianisation of 1930s and spatial distribution of Italians in the 1960s. More people moved in the 1930s also have more Italians in 1961
=> Still after controlling for migration, positive and statistically significant, but the absolute magnitude increase of about 80%
When to use RDD?
Two factors:
1. The treatment of the population must depend on whether an observed variable exceeds a critical value denoted c
- For t to be considered capturing the merit award on academic outcome, it must be that the individuals do not have precise control on the assignment variable.
e. g. should not be manipulated by students, or if the variables were income and taxation, should not be the case that they manipulate their income to receive just income lower than the threshold to pay the lower band of the tax.
Fontana et al (2018) - RDD
The domestic political consequences of the Italian civil war and Nazi occupation during the final two years of WWII
Discontinuity: geographical occupation: municipality just above and below the Gothic line.
Historical background:
Italy was split between troops loyal to Mussolini who fought alongside Germans, and resistance fighters who helped the Allies against germans
Northern Italy was under German occupation - line separating was the Gothic Line
The Treatment for being north of the line is a longer exposure to the Nazi occupation, also means active resistance movements and exposure to Nazi violence.
=> random event that split observations and looked at 50 vs 100 km below and above the Gothic line
Main results:
1. After WWII, municipalities exposed to longer periods of Nazi occupations (above GL) were more likely to vote for Communists - resistance movements. this is because more intense exposure to Nazism let to voters identifying with the political forces that resisted against the enemy
2. Effect end not just temporary, but extremely persistent and enduring lasted until 1992
3. Communist gain was mainly at the expense of the catholic party
=> THE CIVIL WAR CONTRIBUTED TO SHAPE THE POLITICAL IDENTITY OF THE COMMUNIST PARTIES AND GAVE THEM VISIBILITY AND POPULARITY.
Difference-in-Difference (DiD) strategy
The basic idea: evaluate the effects of certain events using pre and post-treatment data.
The treatment can be a change in policy or an exogenous event. (e.g. Becker and Pascali (forth.) - Catholic ban and the Protestant Reformation, Belmonte and Di Lillo (2018) - italicisation of South Tyrol and 1966 reformation for a redistribution of work).
Why diff-in-diff?
- if we compare the outcome in the treated units, we cannot be sure that the difference we attribute is pre-existent
- if we compare the treated units before and after, we cannot be sure that the effect on y is not confounded by other factors => ideally need a treatment and control group, assumed to be similar in everyday except from one change that is the treatment to compare a causal effect.
=> random events are eligible candidates => 2 periods, test control and treatment groups
OLS
Crucial assumption: Cov(x,e)=0
- omitted relevant variables could confound the level of y - unobservables should not explain our results
- simultaneity should not bias an estimator e.g. peoples preferences should not affect the regime - e.g people choose institutions and cannot say about whether people like to live in a communist state when they chose to live in it
=> how to rule these out? exogenous shock, randomisation via historical exogenous events!
to be a valid instrument, z must satisfy two conditions:
1. z must be uncorrelated with the error term (exclusion restriction)
2. z must be correlated with the explanatory variables x (relevant)
=> can use two steps, using the 2SLS
