Road Closure Obstacles and Palestinian Economic Growth A Market Access Approach

Road Closure Obstacles and Palestinian
          Economic Growth
     A Market Access Approach

           Roy van der Weide
              Bob Rijkers
           Brian Blankespoor
            Alexei Abrahams

           ABCDE Conference
             June 21, 2021

                                         1 / 19

How valuable is market access?

    I Hard to assess: Market access generally evolves slowly and
      non-randomly
    I Existing studies exploit variation in:
        I Provision of transportation infrastructure (e.g. Akerman, 2012;
          Banerjee et al., 2012; Baum-Snow et al., 2013; Donaldson,
          2016; Faber, 2014; Jedwab and Moradi, 2014)
        I Placement of borders (e.g. Redding and Sturm, 2008; Brulhart
          et al., 2012)
        I Transportation costs (e.g. Storeygard, 2016; Volpe et al.,
          2014)
    I Do short-term changes in market access impact economic
      growth (proxied by night-time-lights (NTL)) in the West
      Bank?
        I Exploit variation due to mobility restrictions imposed by Israel

                                                                             2 / 19

A map of the West Bank

                         3 / 19

Israel’s Closure Policy

     I Israel deploys obstacles to mobility inside the West Bank
         I Checkpoints, trenches, road bloacks, a barrier wall

     I These serve to protect Israel and Israeli settlements from
       attacks, not to regulate economic performance
       ⇒ Exogeneity

     I Number, intensity and configuration of obstacles changes
       frequently and unexpectedly
       ⇒ spatial and temporal variation in market access

                                                                    4 / 19

Data used

    I Road network data (incl. travel speeds) - from OCHA
    I Road closure obstacles data (incl. wait times) - from OCHA
    I Populated area data (incl. population counts) - from PCBS
    I Night-time-lights data - from NOAA
    I Stock price data - from PEX
    I Fatalities data (both Palestinian and Israeli) - from B’Tselem

                                                                       5 / 19

Optimal (commercial) routes between localities

                                                 6 / 19

Distribution of Travel Times (in minutes)

                                            7 / 19

Measuring market access

                                  X
                         MAit =        Pjt h(Tijt ),
                                   j

   where
    I MAit is market access at locality i and time t
    I h(T ) = T −θ is the “distance function” with distance decay
       parameter θ
    I Tijt is the travel time between origin i and destination j (at
       time t)
    I Pjt is the population in destination j (at time t)
    I The set of destinations comprise the governorate capitals

                                                                       8 / 19

Density of market access (θ = 3)

              .3
              .2
        density
              .1
              0

                   -2          0                      2             4
                                   ln market access

                        2007        2008              2009   2010

                                                                        9 / 19

Market access by locality: 2006 (left) versus 2010 (right)

                                                         10 / 19

Night Time Lights (NTL) - a proxy for economic output

   Figure: Top-coded (left) vs. bottom-coded (right) Night Time Lights
   (2006)

                                                                         11 / 19

NTL versus GDP

            2.4 2.2
       ln(output)
            2
            1.8

                      2000   2002   2004          2006             2008   2010
                                           Year

                                    ln(GDP)              ln(NTL)

                                                                                 12 / 19

Main empirical specification

                ln Yit = β ln MAit + γXit + λi + δt + εit ,
   where:
    I Yit : NTL per capita at location i and time t
    I Xit : ln fatalities (within 5km radius)
    I λi : locality fixed effects
    I δt : time fixed effects

                                                              13 / 19

Identification

     I OLS estimates of β may be biased
         I complementary penalties ⇒ positive bias
         I fixed checkpoints being replaced by flying checkpoints (not
           captured in our database) ⇒ negative bias
         I measurement error in MA ⇒ negative bias

     I Instrument number of checkpoints located in a radius
       between 10 and 25 km from the locality
         I Correlated with MA, but orthogonal to local conditions

                                                                         14 / 19

First stage

              15 / 19

Main results

               16 / 19

Difference-in-Difference Regressions 2006-2010

                                                 17 / 19

Counterfactual growth upon removing all obstacles

                    .4
     log difference in NTL per capita
            .25     .2.3     .35

                                        2004   2006     2008         2010   2012
                                                        year

                                                      θ=1      θ=2
                                                      θ=3      θ=4
                                                      θ=5      θ=6

                                                                                   18 / 19

Conclusion

    I Exploiting quasi-experimental variation in market access
      generated by the deployment of Israeli army obstacles this
      paper demonstrates that short-run fluctuations in market
      access impact local economic performance.

    I Israel’s closure policy reduced GDP per capita in the West
      Bank between 6.0% and 7.5% each year between 2006 and
      2012

    I Preliminary estimates obtained using firm-census data
      suggests that changes in market access primarily impact the
      manufacturing sector

                                                                    19 / 19