Air pollution and agricultural productivity in a developing country

class: center, middle, inverse, title-slide

.title[
# Air pollution and agricultural productivity in a developing country
]
.author[
### Joshua D. Merfeld <br> KDI School and IZA
]
.date[
### 2024-05-10
]

---

/* .remark-slide table{
      width: 100%;
    } */

/* Change the background color to white for shaded rows (even rows) */

.remark-slide thead, .remark-slide tr:nth-child(n) {
      background-color: #A7A9AC;
    }

.remark-slide table {
      background-color: #A7A9AC;
    }
    
    tfoot {
      font-size: 80%;
    }

table{
      border-collapse: collapse;
      border-color: transparent;
      background-color: #A7A9AC;
    }

/* .hljs-github {
      background-image: url("logo.png");
      background-position: bottom left;
      background-size: 10%;
    }

.inverse {
      background-image: url("");
      background-position: bottom left;
      background-size: 10%;
    } */

.title-slide {
      background-image: url("logo_title.png");
      background-position: bottom left;
      background-size: 20%;
    }

.gray {
  color: #7F7F7F;
}

</style>

## This paper
<br>
- Effects of pollution on agricultural productivity in India
  - Previous evidence from gold mines (Aragon and Rud 2016) and workers in California (Graff Zivin and Neidell 2012)
  - Some evidence coal pollution can actually be beneficial (Sanders and Barreca, 2022)<br><br>
- Higher exposure to pollution leads to lower agricultural productivity
  - Exposure defined as wind direction from coal plants towards villages
  - Also instrument for particulate matter 2.5 (PM 2.5)

---

## This paper
<br>
- Larger effects in areas growing more labor-intensive crops
  - Effects through labor?<br><br>
- Compounding effects of shocks
  - Weather shocks + pollution = worse than either on its own<br><br>
- Coal pollution seems worse than other forms of pollution
  - Same increase -> larger effects from coal pollution

---

## Outline

<br>
- General idea and data<br><br>
- Estimating effects of pollution on agricultural productivity<br><br>
- Heterogeneity, including by labor intensities of crops<br><br>
- Overall pollution vs. pollution specifically from coal<br><br>
- Wrapping up

---

## Outline

<br>
- General idea and data<br><br>
- .gray[Estimating effects of pollution on agricultural productivity]<br><br>
- .gray[Heterogeneity, including by labor intensities of crops]<br><br>
- .gray[Overall pollution vs. pollution specifically from coal]<br><br>
- .gray[Wrapping up]

---

## General idea
<br>
- Identify high-pollution locations based on where coal plants open or will open
  - Coal plants open in areas where pollution is already high<br><br>
- Plot wind direction each day
  - Exposure means wind blowing from high-pollution locations to village (within 30km)<br><br>
- Identification comes from changes in exposure to pollution *due to within-village variation* in annual wind direction

---

## Data sources
<table style='NAborder-bottom: 0;border-bottom: 0;border-bottom: 0; font-family: "Arial Narrow", "Source Sans Pro", sans-serif; margin-left: auto; margin-right: auto;' class=" lightable-classic-2">
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> source </th>
   <th style="text-align:center;"> geographic coverage </th>
   <th style="text-align:center;"> temporal coverage </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 2cm; "> shapefile </td>
   <td style="text-align:center;width: 5cm; "> Asher et al. (2021) </td>
   <td style="text-align:center;width: 3cm; "> India </td>
   <td style="text-align:center;width: 3cm; ">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> coal plants </td>
   <td style="text-align:center;width: 5cm; "> Global Energy Monitor </td>
   <td style="text-align:center;width: 3cm; "> global </td>
   <td style="text-align:center;width: 3cm; "> yearly </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> wind </td>
   <td style="text-align:center;width: 5cm; "> NCAR </td>
   <td style="text-align:center;width: 3cm; "> global </td>
   <td style="text-align:center;width: 3cm; "> daily </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> pollution </td>
   <td style="text-align:center;width: 5cm; "> Hammer et al. (2020) </td>
   <td style="text-align:center;width: 3cm; "> global </td>
   <td style="text-align:center;width: 3cm; "> monthly </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> agriculture </td>
   <td style="text-align:center;width: 5cm; "> Gangopadhya et al. (2022) </td>
   <td style="text-align:center;width: 3cm; "> India </td>
   <td style="text-align:center;width: 3cm; "> two seasons/year </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> weather </td>
   <td style="text-align:center;width: 5cm; "> TerraClimate </td>
   <td style="text-align:center;width: 3cm; "> global </td>
   <td style="text-align:center;width: 3cm; "> monthly </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 2cm; "> crops </td>
   <td style="text-align:center;width: 5cm; "> Monfreda et al. (2008) </td>
   <td style="text-align:center;width: 3cm; "> global </td>
   <td style="text-align:center;width: 3cm; "> 2000 (year) </td>
  </tr>
</tbody>
<tfoot><tr><td style="padding: 0; " colspan="100%">
<sup></sup> Global Energy Monitor: globalenergymonitor.org/projects/global-coal-plant-tracker.</td></tr></tfoot>
<tfoot><tr><td style="padding: 0; " colspan="100%">
<sup></sup> NCAR: climatedataguide.ucar.edu/.</td></tr></tfoot>
<tfoot><tr><td style="padding: 0; " colspan="100%">
<sup></sup> TerraClimate: www.climatologylab.org/terraclimate.html</td></tr></tfoot>
</table>

---

## Wind direction - first 100 days of 2010 in Guna district (MP)

---

## Wind direction - first 100 days of 2010 in Guna district (MP)

---

## Where do coal plants open?

<table>
 <thead>
<tr>
<th style="empty-cells: hide;border-bottom:hidden;" colspan="1"></th>
<th style="border-bottom:hidden;padding-bottom:0; padding-left:3px;padding-right:3px;text-align: center; " colspan="2"><div style="border-bottom: 1px solid #ddd; padding-bottom: 5px; ">1991 census</div></th>
<th style="border-bottom:hidden;padding-bottom:0; padding-left:3px;padding-right:3px;text-align: center; " colspan="2"><div style="border-bottom: 1px solid #ddd; padding-bottom: 5px; ">2001 census</div></th>
</tr>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
   <th style="text-align:center;"> (3) </th>
   <th style="text-align:center;"> (4) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> pop (log) </td>
   <td style="text-align:center;width: 4cm; "> -0.107 </td>
   <td style="text-align:center;width: 4cm; "> 0.291*** </td>
   <td style="text-align:center;width: 4cm; "> 0.241*** </td>
   <td style="text-align:center;width: 4cm; "> 0.128 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.097) </td>
   <td style="text-align:center;width: 4cm; "> (0.083) </td>
   <td style="text-align:center;width: 4cm; "> (0.079) </td>
   <td style="text-align:center;width: 4cm; "> (0.080) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> area (log) </td>
   <td style="text-align:center;width: 4cm; "> 0.240*** </td>
   <td style="text-align:center;width: 4cm; "> -0.110 </td>
   <td style="text-align:center;width: 4cm; "> 0.133* </td>
   <td style="text-align:center;width: 4cm; "> 0.119 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.065) </td>
   <td style="text-align:center;width: 4cm; "> (0.088) </td>
   <td style="text-align:center;width: 4cm; "> (0.079) </td>
   <td style="text-align:center;width: 4cm; "> (0.085) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> ag productivity </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 0.092 </td>
   <td style="text-align:center;width: 4cm; "> 0.121 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.070) </td>
   <td style="text-align:center;width: 4cm; "> (0.097) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> pollution </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 0.612*** </td>
   <td style="text-align:center;width: 4cm; "> 0.530** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.201) </td>
   <td style="text-align:center;width: 4cm; "> (0.256) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> has plant? </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 14.842*** </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 13.574*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.568) </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.158) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 209,110 </td>
   <td style="text-align:center;width: 4cm; "> 209,110 </td>
   <td style="text-align:center;width: 4cm; "> 408,913 </td>
   <td style="text-align:center;width: 4cm; "> 415,334 </td>
  </tr>
</tbody>
</table>

---

## Outline

<br>
- .gray[General idea and data]<br><br>
- Estimating effects of pollution on agricultural productivity<br><br>
- .gray[Heterogeneity, including by labor intensities of crops]<br><br>
- .gray[Overall pollution vs. pollution specifically from coal]<br><br>
- .gray[Wrapping up]

---

## Empirical strategy

<br>
First stage

`$$pm_{it} = \alpha_{i} + \gamma_{t} + \phi wind_{it} + \psi f(weather_{it}) + \varepsilon_{it}$$`

Second stage

`$$yield_{it} = \delta_{i} + \eta_{t} + \beta \hat{pm}_{it} + \theta f(weather_{it}) + \upsilon_{it}$$`

- i indexes villages
- t indexes time (year)
- standard errors clustered at village throughout
- common use of stars (one indicates 0.10) throughout

---

## Identification

- Common question: **what if wind direction itself affects yields?**

---

## Wind direction - first 100 days of 2010

---

## Naive regression - yield on pollution

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
   <th style="text-align:center;"> (3) </th>
   <th style="text-align:center;"> (4) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> particulate matter </td>
   <td style="text-align:center;width: 4cm; "> -0.042*** </td>
   <td style="text-align:center;width: 4cm; "> -0.045*** </td>
   <td style="text-align:center;width: 4cm; "> -0.008** </td>
   <td style="text-align:center;width: 4cm; "> -0.041*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> (log PM 2.5) </td>
   <td style="text-align:center;width: 4cm; "> (0.003) </td>
   <td style="text-align:center;width: 4cm; "> (0.003) </td>
   <td style="text-align:center;width: 4cm; "> (0.003) </td>
   <td style="text-align:center;width: 4cm; "> (0.003) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

- Likely endogenous
  - Upward biased? Downard biased?

---

## Reduced form - wind and yield

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
   <th style="text-align:center;"> (3) </th>
   <th style="text-align:center;"> (4) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> wind days </td>
   <td style="text-align:center;width: 4cm; "> -0.0078*** </td>
   <td style="text-align:center;width: 4cm; "> -0.0065*** </td>
   <td style="text-align:center;width: 4cm; "> -0.0051*** </td>
   <td style="text-align:center;width: 4cm; "> -0.0053*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.0007) </td>
   <td style="text-align:center;width: 4cm; "> (0.0007) </td>
   <td style="text-align:center;width: 4cm; "> (0.0007) </td>
   <td style="text-align:center;width: 4cm; "> (0.0007) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## First stage - particulate matter and exposure

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
   <th style="text-align:center;"> (3) </th>
   <th style="text-align:center;"> (4) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> wind </td>
   <td style="text-align:center;width: 4cm; "> 0.008*** </td>
   <td style="text-align:center;width: 4cm; "> 0.008*** </td>
   <td style="text-align:center;width: 4cm; "> 0.008*** </td>
   <td style="text-align:center;width: 4cm; "> 0.008*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.000) </td>
   <td style="text-align:center;width: 4cm; "> (0.000) </td>
   <td style="text-align:center;width: 4cm; "> (0.000) </td>
   <td style="text-align:center;width: 4cm; "> (0.000) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F </td>
   <td style="text-align:center;width: 4cm; "> 1,236 </td>
   <td style="text-align:center;width: 4cm; "> 1,266 </td>
   <td style="text-align:center;width: 4cm; "> 1,205 </td>
   <td style="text-align:center;width: 4cm; "> 1,246 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## Effects of PM on agricultural productivity
<br>
- We are really interested in the effect of _pollution_, not wind<br><br>
- Use wind as instrument? Assumes...
  - Conditional on fixed effects and weather...
  - Wind only affects agriculture through pollution<br><br>
- Reasonable?

---

## Effects of PM on agricultural productivity (IV)

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
   <th style="text-align:center;"> (3) </th>
   <th style="text-align:center;"> (4) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> particulate matter </td>
   <td style="text-align:center;width: 4cm; "> -0.941*** </td>
   <td style="text-align:center;width: 4cm; "> -0.769*** </td>
   <td style="text-align:center;width: 4cm; "> -0.621*** </td>
   <td style="text-align:center;width: 4cm; "> -0.626*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> (log PM 2.5) </td>
   <td style="text-align:center;width: 4cm; "> (0.078) </td>
   <td style="text-align:center;width: 4cm; "> (0.075) </td>
   <td style="text-align:center;width: 4cm; "> (0.080) </td>
   <td style="text-align:center;width: 4cm; "> (0.077) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> No </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (first stage) </td>
   <td style="text-align:center;width: 4cm; "> 1,236 </td>
   <td style="text-align:center;width: 4cm; "> 1,266 </td>
   <td style="text-align:center;width: 4cm; "> 1,205 </td>
   <td style="text-align:center;width: 4cm; "> 1,246 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## Robustness check: leads of wind direction

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> particulate matter </td>
   <td style="text-align:center;width: 5cm; "> -0.033 </td>
   <td style="text-align:center;width: 5cm; ">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> (one-year lead) </td>
   <td style="text-align:center;width: 5cm; "> (0.067) </td>
   <td style="text-align:center;width: 5cm; ">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> particulate matter </td>
   <td style="text-align:center;width: 5cm; ">  </td>
   <td style="text-align:center;width: 5cm; "> -0.070 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> (two-year lead) </td>
   <td style="text-align:center;width: 5cm; ">  </td>
   <td style="text-align:center;width: 5cm; "> (0.052) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (first stage) </td>
   <td style="text-align:center;width: 5cm; "> 592 </td>
   <td style="text-align:center;width: 5cm; "> 783 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 5cm; "> 1,161,265 </td>
   <td style="text-align:center;width: 5cm; "> 1,055,562 </td>
  </tr>
</tbody>
</table>

---

## Heterogeneity

<table>
 <thead>
<tr>
<th style="empty-cells: hide;border-bottom:hidden;" colspan="1"></th>
<th style="border-bottom:hidden;padding-bottom:0; padding-left:3px;padding-right:3px;text-align: center; " colspan="2"><div style="border-bottom: 1px solid #ddd; padding-bottom: 5px; ">wind</div></th>
<th style="border-bottom:hidden;padding-bottom:0; padding-left:3px;padding-right:3px;text-align: center; " colspan="2"><div style="border-bottom: 1px solid #ddd; padding-bottom: 5px; ">yield</div></th>
<th style="empty-cells: hide;border-bottom:hidden;" colspan="1"></th>
</tr>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> &gt;p(50) </th>
   <th style="text-align:center;"> &lt;=p(50) </th>
   <th style="text-align:center;"> &gt;p(50) </th>
   <th style="text-align:center;"> &lt;=p(50) </th>
   <th style="text-align:center;">  </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> PM 2.5 </td>
   <td style="text-align:center;width: 4cm; "> -0.415*** </td>
   <td style="text-align:center;width: 4cm; "> -0.336*** </td>
   <td style="text-align:center;width: 4cm; "> -1.026*** </td>
   <td style="text-align:center;width: 4cm; "> -0.568*** </td>
   <td style="text-align:center;width: 4cm; "> -0.306*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.114) </td>
   <td style="text-align:center;width: 4cm; "> (0.099) </td>
   <td style="text-align:center;width: 4cm; "> (0.060) </td>
   <td style="text-align:center;width: 4cm; "> (0.166) </td>
   <td style="text-align:center;width: 4cm; "> (0.077) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> PM times rain </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 0.119*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> (0.002) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 4cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
   <td style="text-align:center;width: 4cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (1st stage, PM) </td>
   <td style="text-align:center;width: 4cm; "> 597 </td>
   <td style="text-align:center;width: 4cm; "> 482 </td>
   <td style="text-align:center;width: 4cm; "> 675 </td>
   <td style="text-align:center;width: 4cm; "> 428 </td>
   <td style="text-align:center;width: 4cm; "> 674 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (1st stage, PM times rain) </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; ">  </td>
   <td style="text-align:center;width: 4cm; "> 6,393 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 4cm; "> 617,804 </td>
   <td style="text-align:center;width: 4cm; "> 649,118 </td>
   <td style="text-align:center;width: 4cm; "> 634,342 </td>
   <td style="text-align:center;width: 4cm; "> 632,580 </td>
   <td style="text-align:center;width: 4cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## Outline

<br>
- .gray[General idea and data]<br><br>
- .gray[Estimating effects of pollution on agricultural productivity]<br><br>
- Heterogeneity, including by labor intensities of crops<br><br>
- .gray[Overall pollution vs. pollution specifically from coal]<br><br>
- .gray[Wrapping up]

---

## Taking stock
<br>
- We see clear negative effects of pollution on productivity
  - Absolute deviation of exposure is approx. 8 days, meaning 2-3 percent change in productivity
  - Using AD of PM, it's above 20 percent change<br><br>
- Key question: what is the driver?
  - Land?
  - Labor?<br><br>
- Hard to look explicitly at land, so let's look at labor
  - Specifically, effects based on predominant crop

---

## Most common crop by village

---

## How does this help?
<br>
- Different crops have different labor intensities<br><br>
- Rice has the highest (from Michler, 2020):
  - Rice: 1,767 hours per hectare
  - Next highest: cotton at 857.5 hours per hectare

---

## Heterogeneity by rice growing regions

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> particulate matter </td>
   <td style="text-align:center;width: 5cm; "> -0.287** </td>
   <td style="text-align:center;width: 5cm; "> -0.394*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> (log PM 2.5) </td>
   <td style="text-align:center;width: 5cm; "> (0.133) </td>
   <td style="text-align:center;width: 5cm; "> (0.134) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> PM 2.5 times rice </td>
   <td style="text-align:center;width: 5cm; "> -0.338*** </td>
   <td style="text-align:center;width: 5cm; "> -0.205*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 5cm; "> (0.061) </td>
   <td style="text-align:center;width: 5cm; "> (0.060) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 5cm; "> No </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (PM 2.5) </td>
   <td style="text-align:center;width: 5cm; "> 1,369 </td>
   <td style="text-align:center;width: 5cm; "> 1,168 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (PM 2.5 times rice) </td>
   <td style="text-align:center;width: 5cm; "> 2,107 </td>
   <td style="text-align:center;width: 5cm; "> 2,201 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 5cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 5cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## Second possibilitiy: heterogeneity by month of the season
<br>
- Labor demand varies across the season<br><br>
- Pollution may have larger effects during certain months 
  - Between July and October, own farm labor allocation highest in July and October
    (ICRISAT data)<br><br>
- Effects on land should be higher in earlier months, effects on labor relatively higher during harvest

---

## Effects of pollution on productivity by month

---

## Outline

<br>
- .gray[General idea and data]<br><br>
- .gray[Estimating effects of pollution on agricultural productivity]<br><br>
- .gray[Heterogeneity, including by labor intensities of crops]<br><br>
- Overall pollution vs. pollution specifically from coal<br><br>
- .gray[Wrapping up]

---

## Coal plants or overall pollution?
<br>
- I use the location of coal plants to identify high-pollution areas<br><br>
- However, some coal plants open part-way through sample<br><br>
- Differences-in-differences combined with IV
  - Differences in coal pollution, specifically, and other forms of pollution

---

## Coal plants over time

---

## Coal plants or overall pollution?

<table>
 <thead>
  <tr>
   <th style="text-align:left;">   </th>
   <th style="text-align:center;"> (1) </th>
   <th style="text-align:center;"> (2) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 8cm; "> PM 2.5 </td>
   <td style="text-align:center;width: 5cm; "> -0.228*** </td>
   <td style="text-align:center;width: 5cm; "> -0.223*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 5cm; "> (0.066) </td>
   <td style="text-align:center;width: 5cm; "> (0.065) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> PM 2.5 times Coal </td>
   <td style="text-align:center;width: 5cm; "> -0.250*** </td>
   <td style="text-align:center;width: 5cm; "> -0.220*** </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; ">  </td>
   <td style="text-align:center;width: 5cm; "> (0.024) </td>
   <td style="text-align:center;width: 5cm; "> (0.025) </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded) </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> No </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> weather (expanded, bins) </td>
   <td style="text-align:center;width: 5cm; "> No </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; font-weight: bold;"> fixed effects: </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
   <td style="text-align:center;width: 5cm; font-weight: bold;">  </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> year </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> village </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
   <td style="text-align:center;width: 5cm; "> Yes </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (pm) </td>
   <td style="text-align:center;width: 5cm; "> 769 </td>
   <td style="text-align:center;width: 5cm; "> 772 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> F (pm times open) </td>
   <td style="text-align:center;width: 5cm; "> 231 </td>
   <td style="text-align:center;width: 5cm; "> 205 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 8cm; "> observations </td>
   <td style="text-align:center;width: 5cm; "> 1,266,922 </td>
   <td style="text-align:center;width: 5cm; "> 1,266,922 </td>
  </tr>
</tbody>
</table>

---

## Outline

<br>
- .gray[General idea and data]<br><br>
- .gray[Estimating effects of pollution on agricultural productivity]<br><br>
- .gray[Heterogeneity, including by labor intensities of crops]<br><br>
- .gray[Overall pollution vs. pollution specifically from coal]<br><br>
- Wrapping up

---

## Effects on overall agricultural productivity
<br>
- Match changes in pollution across districts with agricultural productivity estimates
  - ICRISAT has district-level estimates from 2001 and 2011
  - Three common crops: rice, wheat, maize<br><br>
- Use most conservative pollution effect estimate: -0.233 from pre-coal estimates
  - My sample overrepresents both pollution and coal pollution

---

## Effects on overall agricultural productivity
<br>
`$$\Delta log(\mathrm{counterfactual})_{d}-0.223\times\Delta log(PM2.5)_{d} = \Delta log(\mathrm{actual})_{d},$$`
<br>
- Use actual observed change in pollution to back out what *would have happened* if pollution had remained at 2001 levels
- Winsorize changes in pollution at 95% (due to some large values)
- Sum total output for entire country in each year

---

## Can't do individual crops...

- I'd like to estimate regressions for each crop individually
  - Better estimates of effects of pollution<br><br>

- However, small sample sizes for many of them:
  - Rice: 759,200
  - Wheat: 376,032
  - Maize: 51,564
  - Cotton (common cash crop): 21,960<br><br>

- Caveat: have to use one estimate for all crops

---

## Change in output and counterfactual, all districts
<br><br>
<table>
 <thead>
  <tr>
   <th style="text-align:left;">  </th>
   <th style="text-align:center;"> 2001 actual output </th>
   <th style="text-align:center;"> 2011 actual output </th>
   <th style="text-align:center;"> 2011 counterfactual </th>
   <th style="text-align:center;"> Difference (pct.) </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;width: 4cm; "> Rice </td>
   <td style="text-align:center;width: 4.6cm; "> 79,962 </td>
   <td style="text-align:center;width: 4.6cm; "> 93,238 </td>
   <td style="text-align:center;width: 4.6cm; "> 95,405 </td>
   <td style="text-align:center;width: 4.6cm; "> 2.3 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 4cm; "> Wheat </td>
   <td style="text-align:center;width: 4.6cm; "> 58,780 </td>
   <td style="text-align:center;width: 4.6cm; "> 93,975 </td>
   <td style="text-align:center;width: 4.6cm; "> 97,520 </td>
   <td style="text-align:center;width: 4.6cm; "> 3.8 </td>
  </tr>
  <tr>
   <td style="text-align:left;width: 4cm; "> Maize </td>
   <td style="text-align:center;width: 4.6cm; "> 11,935 </td>
   <td style="text-align:center;width: 4.6cm; "> 20,897 </td>
   <td style="text-align:center;width: 4.6cm; "> 21,832 </td>
   <td style="text-align:center;width: 4.6cm; "> 4.5 </td>
  </tr>
</tbody>
</table>

---

## Change in pollution
<img src="index_files/figure-html/pollutionincrease-1.png" width="100%" />

---

## Change in output, rice
<img src="index_files/figure-html/counterfactualwheat-1.png" width="100%" />

---

## Conclusion
<br>
- Pollution highest in poorer countries
  - South Asia, in particular<br><br>
- Drivers of pollution are many
  - Coal plants
  - Transportation
  - Industry
  - Agriculture
  - etc.<br><br>
- We already have plenty of evidence on effects of pollution on health

---

## Conclusion
<br>
- Paper presents evidence that air pollution leads to large decreases in agricultural productivity
  - Identification relies on within-village changes across seasons<br><br>
- Air pollution worsened by around 15% from 1998 to 2019
  - Conservative results indicate decreases of between 3 and 5 percent in agricultural productivity <br><br>
- Key finding: effects larger where crops are more labor intensive

---

## Conclusion
<br>
- Overall cost/benefit way beyond scope of paper<br><br>
- Instead, results point to importance of the location of pollution sources
  - Downwind villages suffer<br><br>
- 30km radius indicates relatively long-range effects

---

class: center, middle

<font size = "40"> Thank you! </font>

[https://joshmerfeld.github.io](https://joshmerfeld.github.io) <br>
Twitter: [@Josh\_Merfeld](twitter.com/Josh_Merfeld)