This is the third in our series of posts by students on the job market this year.
Economists tend to believe that travel and trade costs reduce welfare. Trade papers like Irwin (2005), Redding & Sturm (2008), Storeygard (2014), and Etkes & Zimring (2014) draw on evidence from the United States, West Germany, sub-Saharan Africa, and the Gaza Strip to support this idea. One might reasonably expect, therefore, that the welfare of Palestinian commuters declined during the Second Palestinian Uprising (2000-2007), when the Israeli army deployed hundreds of roadblocks and checkpoints along the West Bank’s internal road network in order to defend Israeli civilian settlements. Although these obstacles were intended to deter and intercept militants, they had the unintended consequence of delaying Palestinian civilian travel between Palestinian towns, and from Palestinian towns to Israel (B’Tselem (2007), World Bank (2007)). Two World Bank working papers (Cali & Miaari (2014), van der Weide et al (2014)) take advantage of this ‘natural experiment’ to study the effects of travel costs on commuters’ welfare, finding that economic outcomes of Palestinians declined in the face of obstacle deployment. My job market paper, however, finds a very different result: while obstacles reduced the welfare of laborers in some towns, laborers from other towns actually benefited from obstacles. The salient outcome of obstacle deployment was not welfare reduction, but rather welfare inequality.
How could any Palestinian laborers have benefited from obstacles? I develop a commuting model to formalize a simple insight: a laborer’s welfare declines as travel costs increase along the road to his job; but his welfare rises when travel costs increase for other laborers, since reduced labor supply to firms induces them to raise wages. Stated another way, laborers are competing with each other for jobs, and competition bids wages down. When some laborers face higher travel costs to reach jobs, they abandon those jobs in favor of unemployment benefits. Their absence from the marketplace is a boon for the remaining laborers, who enjoy increased wages. Travel costs create winners and losers; one laborer’s loss is another’s gain.
Obtaining 2007 census data of 485 Palestinian localities, I ask how localities’ post-uprising employment rates responded to obstruction from jobs; and protection from competing labor. I georeference and digitize UN maps showing obstacle locations along roads connecting localities during 2003-2007. The same exact obstacles are responsible for both countervailing effects, but I tease them apart using variation in pre-uprising jobs shares and pre-uprising labor shares of each locality. Intuitively, residential towns suffered more obstruction than protection; vise versa for commercial towns. I obtain pre-uprising labor force shares from the 1997 population census. For job shares, I use the 2004 (mid-uprising) firm census. Concerned that these shares may be endogenous to treatment, I establish that nighttime light emissions of Palestinian towns, as recorded by US-DMSP satellites, are a credible proxy for localities’ job counts. I then impute pre-uprising job shares using pre-uprising nighttime lights emissions over Palestinian localities. Qualitatively, results using these radiance-imputed pre-uprising job shares end up being very similar to results using the mid-uprising job shares. Check out my second dissertation paper to see how I fix optical distortions in these imagery.
Israeli civilian settlements were scattered throughout the West Bank during the uprising, and obstacles were deployed in order to protect those settlements (B’Tselem (2007), World Bank (2007)). I find that settlement proximity to Palestinian commuter routes is an excellent predictor of how many obstacles each Palestinian town faced, and uncorrelated with other characteristics of these towns relevant to economic outcomes. I use proximity of settlements to commuter routes to make instruments for ‘obstruction’ and ‘protection’.
I perform 2SLS regressions to estimate the countervailing effects of obstacles on 2007 (post-uprising) employment rates of 485 census localities. Integrating over locality labor force sizes, the aggregate effect of obstacles was to reduce total employment of the economy by just .32 percentage points, with the protective effect (3.82 percentage points) importantly mitigating the obstructive effect (-4.14 percentage points). Indeed, projecting the net effect fitted values, some 174 of 485 Palestinian census localities enjoyed net gains to employment as a result of obstacles. These localities (see orange- and red-colored circles in Figure 1) were predominantly in the north-middle of the West Bank, between the major economic centers of Ramallah and Nablus. In the pre-uprising era, the flow of commuters would have tended to be from periphery to core (toward jobs). In that case, when travel costs increased, laborers dwelling in the core would have been largely impervious to obstacles’ obstructive effects because they were never traveling far to work. Moreover, as the inflow of peripheral laborers was stymied, core residents were ideally situated to benefit from increased wages.
Limitations of the study:
Short run v. Long run:
My results are only valid in the short run. In the long run, welfare inequality disappears because firms and laborers relocate. In the developing world, however, the short run can be pretty long. Topalova (2010), Munshi & Rosenzweig (2013), and Bryan et al (2014) all find persistent inequality due to factor immobility between city and rural areas in India and Bangladesh. I find the same is true in the West Bank: only 1.6% of 2007 labor force census respondents reported moving between 2000-2007 for job-related reasons. Having established nighttime light emissions over Palestinian localities as a valid proxy for job counts, I use pre- and post-uprising lights to show that localities' proximity to jobs changed minimally over the time period.
Homogeneous v. Heterogeneous labor:
My model assumes labor is homogeneous. This makes sense for my context, and for many urban areas of developing countries, where low-skill laborers work menial jobs and are easily replaced if travel costs delay them. My results do not extend well to commuter economies with specialized labor.
What about wages?
I do not observe wages. In my model, laborers discover one job opportunity per time period, so employment rate changes are a sufficient statistic for welfare changes. If, however, laborers can switch to backup job opportunities, then wages are better at tracking welfare changes.
Travel costs in a commuter economy can have heterogeneous consequences, creating winners and losers. If the distribution of jobs is concentrated in just a few, core locations, then this heterogeneity will take on a clear spatial pattern: the winners will tend to be residents nearest to core locations, while the losers will tend to be residents dwelling in the periphery. This core-periphery inequality should factor into municipal decision making about transport infrastructure.
Policy takeaway for the Israel-Palestine Conflict:
Obstacles generated inequality within the Palestinian economy, leaving core areas (Ramallah-Nablus) better off and peripheral areas worse off (Figure 1). Anecdotally supporting my story, residents of Ramallah sometimes resentfully refer to non-Ramallah Palestinians flowing in to town for jobs as ‘Thailandiyas’, alluding to a similar labor market event where Thai guest workers immigrated to Israel and seized many low-skill jobs Palestinians had previously monopolized (Friedberg and Sauer (2003)). Going forward, we may expect that laborers from the West Bank’s periphery will relocate to the core to improve their welfare (already happening in Ramallah). Periphery-to-core migration in the West Bank is cause for great concern: In the ongoing context of Israeli settlement expansion (World Bank (2013)) and recent land grabs (NY Times article), depopulation of peripheral areas invites annexation by Israel, further eclipsing the possibility of a two-state solution.
Alexei Abrahams is a PhD student at the Department of Economics and Population Studies & Training Center, Brown University.