Works in progress
With M Moro (Stirling) and E Moore (Stirling)
The impact of University reopenings on COVID-19 cases in Scotland
This paper estimates the impact of University reopenings in Scotland in Autumn 2020 on COVID-19 cases in Scottish neighbourhoods. We geolocate all student halls in Scotland, and merge this data with neighbourhood-level case data. We employ a local differences-in-differences strategy and tackle two research questions. First, we ask what was the impact of the start of semester on cases in the student neighbourhoods? Next, we turn our attention to the spillover of cases in the nearby communities to student neighbourhoods.
University semester start dates in Scotland are staggered over the month of September, and we deal with this by focusing on each start cluster, as well as implementing the Callaway and Sant’Anna (2020) estimator. We find a substantial and persistent increase in cases in areas containing halls and evidence of persistent spillovers. These effects are linked to the group of Universities that started on 14th September, which include large
Universities located in the major urban areas. The cases began to rise on 21st September, with 100 extra cases per 100,000, and peaked a week later with 400 per 100,000, after which they started declining, but persist until the Autumn tightening of coronavirus restrictions bit in November, two months after the restrictions were enacted. Our results invite a re-think of how close contact activities may safely resume.
Keywords Covid Economics, University reopenings, cases, Public Health
JEL Codes I10, I18, I28
With A Schwarz (PIK Potsdam) E Egger (UNU-WIDER) and C Poggi (AFD) and M Moro (Stirling)
The unequal effect of pollution exposure on labour supply across gender
This paper studies the effects of air pollution on labour supply by merging information on hours worked by millions of people living in Mexico City with detailed pollution data over the period 2005 to 2010. When pollution exceeds particular thresholds Mexico City activates the ”Environmental Contingency Program” tiers to a Pre-contingency or a more severe Contingency level. We exploit these thresholds and adopt a regression discontinuity design to document a reduction in the hours worked when a (pre)contingency is called. We then study how daily working hours evolve around a (pre)contingency and shows how labour supply responds before and after its activation. Further, we supplement this evidence with information on pollution from measurement stations across the city. We link this data with a labour force survey that is representative survey at city level and explore the effects of exposure of pollution at PM2.5. We find evidence that contemporaneous pollution exposure at moderate levels reduce labour supply, whilst there are dramatic reductions at high levels of pollution with heterogeneous effects by formality status. Moreover, pollution seems to decrease working hours even in non-emergency times with differential effects by gender. There is an unequal gendered labour response when exposed to pollution and this effect is heterogeneous across employment type. Where most female and make formal workers are able to reduce their hours of work, informal female workers may have no escape options. Female informal workers have the highest increase in minutes of work during the peak pollution days. We inspect lags pollution exposure to corroborate our findings with respect to a delayed response to the exposure event.
Presentation (last updated 04/19)
(Major update coming soon)
Striking a Deal? Strike threats and the business cycle in Mexico
It is well documented that unions have effects on wages and non-wage benefits, these gains are obtained through collective bargaining. The theoretical literature motivates these gains as the products of successful bargains achieved using the threat of a strike as the bargaining signal. This paper provides a novel exploration of the bargaining signalling mechanism between unions and employers by exploiting administrative data on all strike threats in Mexico in the private sector 1991-2012. I explore the relationship between strike threats and the business cycle for a 20 year period and confirm their role in wage bargaining for workers.
What have Mexican unions done to wages over the last decade?
With Willem Sas (Stirling)
Institutions in the Fast Lane? Strike Petitions and the Electoral Cycle in Mexico
This paper exploits Mexican administrative data on all strike threats between 1991–2012, a period of political institutional change. This paper asks: Are strike threats partially caused by the political cycle? We develop a political economy model of union influence and find that when electoral institutions become more democratic, and political parties can count to a lesser extent on a guaranteed support base, the importance of relying in the unions to win
elections will increase. We validate these findings using municipal electoral data, where we employ a sharp regression discontinuity approach, and find a causal effect from close elections of right- and left-wing mayors on strike threats two years after an election. Narrow electoral victories of the right-wing (left-) party increase the number of strike threats by 1.056 (1.456)
per 10,000 of the municipal population two years after the election. This finding is robust to alternate specifications. We suggest that threats may be misused for campaigning in upcoming elections. To test this hypothesis a differences-in-differences model is employed to estimate changes in electoral turnout in narrow win municipalities. We find that electoral turnout is
stimulated by strike threats, in the context of tight electoral rules surrounding campaigning, these findings may be interpreted confirmation of illegal campaigning.
JEL Codes: C23, D72, J52, P48
PDF (last updated 03/19)
With A Newell (Sussex)
Joining the dots and fitting curves: A practitioner’s guide to estimating income inequality using group data
One of the challenges of working with historical data relate to the passage of time. Often, the microdata of surveys no longer survives. This paper is a guide on how to estimate income inequality from tabulated results that may survive in reports. We consider various data structures and propose parametric estimation methods which overcome the lack of individual returns. Microsimulations using the 1953-54 UK MoL Survey (Gazeley et al., 2015) and the 1853 Survey of Belgian workers (Dupéctiaux, 1855), for which individual level data exist, are employed to demonstrate the performance of these methods once the data have been collapsed into groups on modern stratified data and snowball samples, respectively. (PDF by request)
with I Gazeley (LSE), C Lanata Briones(Warwick), A Newell (Sussex), K Reynolds (Brighton) and R Holmes (Sussex)
“Inequality among European working households 1890–1960,” IZA DP 11355 February 2018. PDF here
In this article we map, for the first time, the time-path of the size distribution of income among working class households in Western Europe, 1890-1960. To do this we exploit data extracted from a large number of newly digitised household expenditure surveys. Many are not representative of the population, or even of their target-subpopulation, as methods of social investigation were initially primitive, though rapidly evolving over this period. We overcome the consequent problem of comparability by exploiting our knowledge of the methods used by early social investigators to estimate of the scale of known biases. For some we have the original household data, but in most cases we have tables by income group. One by-product of this work is an evaluation of the range of estimation methods for distributional statistics from these historical tables of grouped data. Our central finding is that inequality among working households does not follow the general downward trend in inequality for the early part of the century found in labour share and top income studies.
Contrary to Kuznets’ prediction, our evidence suggests that on average income inequality among European working households remained stable for three generations from the late nineteenth century onwards
JEL Classification: N33, N34, O15
Keywords: inequality, working households, Europe, 20th century
“Latin American Household Budget Surveys 1913-1970 and What They Tell Us about Economic Inequality among Households,” IZA DP 11430 March 2018. PDF here
with I Gazeley (LSE), A Newell (Sussex), and K Reynolds (Brighton)
“Household survey-based evidence on economic inequality in Britain, 1937-61” revisions requested The Economic History Review
A previous (and substantially different version previously circulated under“What Really Happened to British Inequality in the Early 20th Century? Evidence from National Household Expenditure Surveys 1890–1961,” IZA DP 11071 October 2017. PDF here )
We estimate household income inequality for the UK 1937- 1961. Firstly, we investigate the change in inequality between 1937/8 and 1953/4. Tax unit data show a fall of seven Gini points, due to a collapse the share of the top 10%, but no change among lower incomes. Our household data also cannot reject the idea of no movement in inequality below the highest incomes. Secondly, we compare data from 1953/4 to the Family Expenditure Survey for 1961. We find that by 1961, inequality was higher than 1953/4. Investigating that rise reveals a role the emergence of small, low income, mostly non-working households, in shaping the change between 1953/4 and 1961. For instance, there was a doubling in the share of retired households over the period, a group who exhibit greater inequality than working households. Interestingly, the impact on overall economic inequality of the rise of this social group is partially offset by the fact that they are, by and large, both smaller and poorer. This raises the covariance of income and household size, which offsets the impact on inequality of the rises in the variances of income and household size.