Website: hosamibrahim.com
#EconSky #EconJobMarket #DevEcon #AgEcon #Mining #Agriculture

🧩 Mechanism

Evidence points to labor-market competition, not pollution as the short- to medium-run primary mechanism.

Contrast with Ghana (Aragón & Rud) likely reflects timing & measurement. [13/13]

Output falls as inputs fall. Conditional on labor & land (Cobb–Douglas), no systematic decline in yields or output → TFP stable in short run. Soil/rooting indices show no early deterioration (sometimes mild improvements consistent with fallowing).
[12/13]

⏱️ Dynamics across countries

Near-mine farms (0–15 km) show persistent family-labor declines from t+2 onward; hired labor flat; total labor mirrors family labor. Effects attenuate with distance; pre-trends small & jointly insignificant.
[11/13]

Within 15 km: family labor −12.5 days (~16%, p<0.01); hired labor ~0 → total labor −11.4 days (p<0.01). Planted area falls, but yields/TFP don’t once conditioning on inputs. [10/13]

📐 Exposure & ID

Treatment at the enumeration area level defined using distance bins (0–15, 15–30, 30–45 km) between an EA and a mine.

Never-exposed EAs (located more than 45 km away from any mine) are untreated; country-year FEs; EA SE clustering. [9/13]

Design 1) Tanzania (2011–2013): 2x2 TWFE DiD around a 2012 gold-mine opening (panel plots)

Design 2) ETH/TZA/UGA (2009–2022): staggered event study using repeated cross-sections of plots in the 3 countries

Data: LSMS-ISA (plots) + MinEx [8/13]

To disentangle pollution and input competition, I estimate ATET of being close to a mine on:

▶ Farm labor days and productivity.
▶ Planted area, harvest, and yield.
▶ Soil and plant rooting conditions (suggestive evidence).

..at different periods of exposure time. [7/13]

Below is an example of the VMPL for three hypothetical farms located at three different distance bins from a mine:

(a) before a mine opens,
(b) right after a mine opens, and
(c) 3-4 years after a mine opens. [6/13]

I ask the question: Are early agricultural changes driven by workers leaving farms or by pollution harming soils and plants?

I argue both mechanisms run on different clocks: labor shocks are immediate; pollution accumulates slowly. [5/13]

Another example in Tanzania's Geita region (a major gold mining region), showing changes in NDVI in the Bukombe-Mbogwe Forest Reserve:

(a) 1984, (b) 2002, (c) 2020 (Pancrace et al., 2022) [4/13]

An example of deforestation and loss of vegetation in Tanzania's Singida region (known for gold mining):

Gold mining started in 2004 in Singida, see Lameck et al., 2025 [3/13]

Mines (red) increasingly overlap with agriculture (green). Example below from Tanzania using LSMS-ISA and MinEx data. [2/13]