lp.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Estimating the macroeconomic effect of conflicts with local projections"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "***Author:*** N. Julitz\n",
    "\n",
    "***Data:*** (UCDP) Georeferenced Event Dataset, GDP per capita (IMF), World Population Data (Kaggle)\n",
    "\n",
    "\n",
    "***Further info:***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### First, we need to import the relevant libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from linearmodels.panel import PanelOLS\n",
    "from linearmodels.panel import compare\n",
    "import statsmodels.api as sm\n",
    "import datetime as datetime"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Then, we need to import the datasets we want to use"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {},
   "outputs": [],
   "source": [
    "ucdp = pd.read_csv(\"UCDP.csv\", low_memory=False)\n",
    "\n",
    "gdp = pd.read_excel(\"GDP.xlsx\")\n",
    "\n",
    "population = pd.read_csv(\"population.csv\", sep=\",\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making the ucdp-dataset ready for statistical analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 349733 entries, 0 to 349732\n",
      "Data columns (total 49 columns):\n",
      " #   Column             Non-Null Count   Dtype  \n",
      "---  ------             --------------   -----  \n",
      " 0   id                 349733 non-null  int64  \n",
      " 1   relid              349733 non-null  object \n",
      " 2   year               349733 non-null  int64  \n",
      " 3   active_year        349733 non-null  int64  \n",
      " 4   code_status        349733 non-null  object \n",
      " 5   type_of_violence   349733 non-null  int64  \n",
      " 6   conflict_dset_id   349733 non-null  int64  \n",
      " 7   conflict_new_id    349733 non-null  int64  \n",
      " 8   conflict_name      349733 non-null  object \n",
      " 9   dyad_dset_id       349733 non-null  int64  \n",
      " 10  dyad_new_id        349733 non-null  int64  \n",
      " 11  dyad_name          349733 non-null  object \n",
      " 12  side_a_dset_id     349733 non-null  int64  \n",
      " 13  side_a_new_id      349733 non-null  int64  \n",
      " 14  side_a             349733 non-null  object \n",
      " 15  side_b_dset_id     349733 non-null  int64  \n",
      " 16  side_b_new_id      349733 non-null  int64  \n",
      " 17  side_b             349733 non-null  object \n",
      " 18  number_of_sources  349733 non-null  int64  \n",
      " 19  source_article     349733 non-null  object \n",
      " 20  source_office      251211 non-null  object \n",
      " 21  source_date        251211 non-null  object \n",
      " 22  source_headline    251205 non-null  object \n",
      " 23  source_original    306273 non-null  object \n",
      " 24  where_prec         349733 non-null  int64  \n",
      " 25  where_coordinates  349733 non-null  object \n",
      " 26  where_description  341001 non-null  object \n",
      " 27  adm_1              331054 non-null  object \n",
      " 28  adm_2              280734 non-null  object \n",
      " 29  latitude           349733 non-null  float64\n",
      " 30  longitude          349733 non-null  float64\n",
      " 31  geom_wkt           349733 non-null  object \n",
      " 32  priogrid_gid       349733 non-null  int64  \n",
      " 33  country            349733 non-null  object \n",
      " 34  country_id         349733 non-null  int64  \n",
      " 35  region             349733 non-null  object \n",
      " 36  event_clarity      349733 non-null  int64  \n",
      " 37  date_prec          349733 non-null  int64  \n",
      " 38  date_start         349733 non-null  object \n",
      " 39  date_end           349733 non-null  object \n",
      " 40  deaths_a           349733 non-null  int64  \n",
      " 41  deaths_b           349733 non-null  int64  \n",
      " 42  deaths_civilians   349733 non-null  int64  \n",
      " 43  deaths_unknown     349733 non-null  int64  \n",
      " 44  best               349733 non-null  int64  \n",
      " 45  high               349733 non-null  int64  \n",
      " 46  low                349733 non-null  int64  \n",
      " 47  gwnoa              265919 non-null  object \n",
      " 48  gwnob              15910 non-null   float64\n",
      "dtypes: float64(3), int64(25), object(21)\n",
      "memory usage: 130.7+ MB\n"
     ]
    }
   ],
   "source": [
    "ucdp.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "202"
      ]
     },
     "execution_count": 261,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ucdp[\"total_deaths\"] = ucdp.deaths_a + ucdp.deaths_b + ucdp.deaths_civilians # make a column for the total deaths\n",
    "\n",
    "ucdp[\"conflict_dummy\"] = ucdp[\"total_deaths\"].apply(lambda x: 1 if x > 1000 else 0) # all conflicts, that bypass a threshold of 1,000 deaths, are coded to 1 and all others are coded to 0\n",
    "\n",
    "len(ucdp[ucdp[\"conflict_dummy\"] == 1]) # the total threshold identifies 202 conflicts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['id', 'relid', 'year', 'active_year', 'code_status', 'type_of_violence',\n",
       "       'conflict_dset_id', 'conflict_new_id', 'conflict_name', 'dyad_dset_id',\n",
       "       'dyad_new_id', 'dyad_name', 'side_a_dset_id', 'side_a_new_id', 'side_a',\n",
       "       'side_b_dset_id', 'side_b_new_id', 'side_b', 'number_of_sources',\n",
       "       'source_article', 'source_office', 'source_date', 'source_headline',\n",
       "       'source_original', 'where_prec', 'where_coordinates',\n",
       "       'where_description', 'adm_1', 'adm_2', 'latitude', 'longitude',\n",
       "       'geom_wkt', 'priogrid_gid', 'country', 'country_id', 'region',\n",
       "       'event_clarity', 'date_prec', 'date_start', 'date_end', 'deaths_a',\n",
       "       'deaths_b', 'deaths_civilians', 'deaths_unknown', 'best', 'high', 'low',\n",
       "       'gwnoa', 'gwnob', 'total_deaths', 'conflict_dummy'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 262,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ucdp.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert all dates to datetime-objects\n",
    "ucdp[\"date_start\"] = pd.to_datetime(ucdp[\"date_start\"])\n",
    "ucdp[\"date_end\"] = pd.to_datetime(ucdp[\"date_end\"])\n",
    "\n",
    "# add a new DataFrame with years from 1988-2023\n",
    "ucdp = ucdp[(ucdp[\"year\"] >= 1988) & (ucdp[\"year\"] <= 2023)].copy()\n",
    "\n",
    "# dropping all columns not needed\n",
    "conflict = ucdp.drop(ucdp.columns[[0, 1, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 34, 35, 36, 37, 44, 45, 46, 47, 48]], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying the definition for conflict-onset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [],
   "source": [
    "# we need to classify a new conflict onset after 4 years of not reaching the threshold\n",
    "\n",
    "# sort the rows after country and year (descending)\n",
    "conflict.sort_values(by=[\"country\", \"year\", \"date_start\"], ascending=False, ignore_index=True, inplace=True)\n",
    "\n",
    "# grouping the dataset after country and years and sum up the deaths for each year\n",
    "conflict_grouped = conflict.groupby(by=[\"country\", \"year\"]).total_deaths.sum().unstack().copy()\n",
    "conflict_grouped = conflict_grouped.reset_index().copy()\n",
    "conflict_grouped = conflict_grouped.melt(id_vars=\"country\").copy()\n",
    "conflict_grouped = conflict_grouped.rename(columns={\"value\": \"total_deaths\"}).copy()\n",
    "conflict_grouped.sort_values(by=[\"country\", \"year\"], ascending=False, ignore_index=True, inplace=True)\n",
    "\n",
    "# by grouping, the dataframe was changed into wide-format, which caused all countries to appear in every year column, even though not every country has a conflict in every year\n",
    "# we need to change that by dropping all NaN-values\n",
    "conflict_grouped = conflict_grouped.dropna().copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [],
   "source": [
    "# problem: for the application of the conflict-onset-definition, we need to make sure, that there is no years are left out in the countries\n",
    "# otherwise, the application will set years as conflict-onset, which are not\n",
    "\n",
    "all_years = pd.DataFrame()\n",
    "\n",
    "for country in conflict_grouped['country'].unique():\n",
    "    country_df = conflict_grouped[conflict_grouped['country'] == country]\n",
    "    min_year = country_df['year'].min()\n",
    "    max_year = country_df['year'].max()\n",
    "    full_years = pd.DataFrame({'year': range(min_year, max_year + 1)})\n",
    "    full_years['country'] = country\n",
    "    all_years = pd.concat([all_years, full_years])\n",
    "\n",
    "conflict_grouped = pd.merge(all_years, conflict_grouped, on=['country', 'year'], how='left')\n",
    "\n",
    "conflict_grouped[\"total_deaths\"] = conflict_grouped[\"total_deaths\"].astype(float).fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {},
   "outputs": [],
   "source": [
    "# again, we need to apply the 1000 deaths threshold\n",
    "conflict_grouped[\"conflict_dummy\"] = conflict_grouped[\"total_deaths\"].apply(lambda x: 1 if x > 1000 else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/rx/hnk5_pq93gs_md650fn70c6m0000gn/T/ipykernel_33525/1398739182.py:19: DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns. This behavior is deprecated, and in a future version of pandas the grouping columns will be excluded from the operation. Either pass `include_groups=False` to exclude the groupings or explicitly select the grouping columns after groupby to silence this warning.\n",
      "  conflict_grouped['conflict_onset'] = country_grouped.apply(lambda x: check_condition(x, target_value)).reset_index(level=0, drop=True)\n"
     ]
    }
   ],
   "source": [
    "# applying the definition for conflict-onset and create a new column for that\n",
    "\n",
    "target_value = 0\n",
    "\n",
    "country_grouped = conflict_grouped.groupby('country')\n",
    "\n",
    "def check_condition(group, target_value):\n",
    "    \n",
    "    shifted_0 = group['conflict_dummy']\n",
    "    shifted_1 = group['conflict_dummy'].shift(1)\n",
    "    shifted_2 = group['conflict_dummy'].shift(2)\n",
    "    shifted_3 = group['conflict_dummy'].shift(3)\n",
    "    shifted_4 = group['conflict_dummy'].shift(4)\n",
    "\n",
    "    condition = (shifted_0 == 1) & (shifted_1 == target_value) & (shifted_2 == target_value) & (shifted_3 == target_value) & (shifted_4 == target_value)\n",
    "    return condition\n",
    "\n",
    "\n",
    "conflict_grouped['conflict_onset'] = country_grouped.apply(lambda x: check_condition(x, target_value)).reset_index(level=0, drop=True)\n",
    "\n",
    "conflict_grouped[\"conflict_onset\"] = conflict_grouped[\"conflict_onset\"].astype(str).copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {},
   "outputs": [],
   "source": [
    "conflict_grouped.loc[conflict_grouped[\"conflict_onset\"] == \"True\", \"conflict_onset\"] = 1\n",
    "conflict_grouped.loc[conflict_grouped[\"conflict_onset\"] == \"False\", \"conflict_onset\"] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [],
   "source": [
    "conflict = conflict_grouped"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making the gdp-per-capita-dataset ready for statistical analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {},
   "outputs": [],
   "source": [
    "# data is in us-dollar per capita\n",
    "gdp_melted = pd.melt(gdp, id_vars=[\"country\"]).copy() # change the dataframe from wide-format to long-format\n",
    "\n",
    "gdp_melted.rename(columns={\"variable\": \"year\", \"value\": \"gdp\"}, inplace=True) # renamed columns\n",
    "\n",
    "# add a new dataframe with years from 1988-2023\n",
    "gdp_final = gdp_melted[(gdp_melted[\"year\"] >= 1988) & (gdp_melted[\"year\"] <= 2023)].copy()\n",
    "gdp_final.reset_index(inplace=True)\n",
    "del(gdp_final[\"index\"])\n",
    "\n",
    "# changing the data-format of columns year and gdp\n",
    "gdp_final.loc[gdp_final[\"gdp\"] == \"no data\", \"gdp\"] = np.nan\n",
    "gdp_final = gdp_final[gdp_final[\"country\"] != \"©IMF, 2024\"].copy()\n",
    "gdp_final = gdp_final[~gdp_final[\"country\"].isna()] # deleting all country-entries which have missing values\n",
    "\n",
    "# imputation of missing values in gdp column with mean value of gdp per capita for the specific country\n",
    "mean_gdp = gdp_final.groupby(by=\"country\").gdp.mean().copy()\n",
    "\n",
    "gdp_merged = pd.merge(left=gdp_final, right=mean_gdp, on=\"country\", how=\"left\").copy() # left-join both dataframes\n",
    "\n",
    "gdp_merged['gdp_x'] = gdp_merged['gdp_x'].combine_first(gdp_merged['gdp_y']) # replace all missing values with the mean value of its country\n",
    "\n",
    "gdp_merged.rename(columns={\"gdp_x\": \"gdp\", \"gdp_y\": \"gdp_mean\"}, inplace=True) # renaming columns\n",
    "\n",
    "gdp_merged[\"gdp\"] = gdp_merged[\"gdp\"].astype(float).copy()\n",
    "\n",
    "gdp_merged.sort_values(by=[\"country\", \"year\"], ascending=False, ignore_index=True, inplace=True)\n",
    "\n",
    "gdp_merged.drop(columns=[\"gdp_mean\"], inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making the population-dataset ready for statistical analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "# dropping and renaming some columns\n",
    "population.drop(columns=[\"rank\", \"cca3\", \"continent\", \"density (km²)\", \"area (km²)\", \"growth rate\", \"world percentage\"], inplace=True)\n",
    "\n",
    "population.rename(columns={\"2023 population\": \"2023\", \"2022 population\": \"2022\", \"2020 population\": \"2020\", \"2015 population\": \"2015\", \"2010 population\": \"2010\", \"2000 population\": \"2000\", \"1990 population\": \"1990\", \"1980 population\": \"1980\", \"1970 population\": \"1970\"}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [],
   "source": [
    "population = population.melt(id_vars=\"country\")\n",
    "\n",
    "population.rename(columns={\"variable\": \"year\", \"value\": \"population\"}, inplace=True)\n",
    "\n",
    "population.sort_values(by=[\"country\", \"year\"], ascending=False, ignore_index=True, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Now, we need to merge all three datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [],
   "source": [
    "# use a left-join on the conflict-dataset, since this is the data, we are dependent on\n",
    "\n",
    "joined_df = pd.merge(conflict, gdp_merged, on=[\"country\", \"year\"], how=\"left\")\n",
    "\n",
    "joined_df.dropna(inplace=True, ignore_index=True)\n",
    "\n",
    "joined_df.sort_values(by=[\"country\", \"year\"], ascending=True, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_rows_with_following(df, binary_col, entity_col, n_following=10):\n",
    "    # Create an empty list to store the indices of rows to keep\n",
    "    rows_to_keep = []\n",
    "    \n",
    "    # Iterate over unique entities\n",
    "    for entity in df[entity_col].unique():\n",
    "        # Filter rows for the current entity\n",
    "        entity_df = df[df[entity_col] == entity]\n",
    "        \n",
    "        # Get the indices where the binary column is 1\n",
    "        indices = entity_df.index[entity_df[binary_col] == 1]\n",
    "        \n",
    "        for idx in indices:\n",
    "            # Calculate the range of rows to include\n",
    "            end_idx = min(idx + n_following, entity_df.index[-1] + 1)\n",
    "            rows_to_keep.extend(range(idx, end_idx))\n",
    "    \n",
    "    # Remove duplicates and keep the rows in the original DataFrame\n",
    "    rows_to_keep = sorted(set(rows_to_keep))\n",
    "    new_df = df.loc[rows_to_keep]\n",
    "    \n",
    "    return new_df\n",
    "\n",
    "# Example usage\n",
    "new_df = extract_rows_with_following(joined_df, 'conflict_onset', 'country', n_following=10)\n",
    "joined_df = new_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {},
   "outputs": [],
   "source": [
    "joined_df.set_index([\"country\", \"year\"], inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>total_deaths</th>\n",
       "      <th>conflict_dummy</th>\n",
       "      <th>conflict_onset</th>\n",
       "      <th>gdp</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>country</th>\n",
       "      <th>year</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"2\" valign=\"top\">Ukraine</th>\n",
       "      <th>2022</th>\n",
       "      <td>73820.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4582.609000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2023</th>\n",
       "      <td>70809.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>5337.491000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"10\" valign=\"top\">Uganda</th>\n",
       "      <th>1996</th>\n",
       "      <td>1743.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>401.413000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1997</th>\n",
       "      <td>1313.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>423.918000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998</th>\n",
       "      <td>1202.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>402.455000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1999</th>\n",
       "      <td>820.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>370.515000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000</th>\n",
       "      <td>1157.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>355.466000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2001</th>\n",
       "      <td>212.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>347.812000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2002</th>\n",
       "      <td>1309.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>361.953000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2003</th>\n",
       "      <td>1592.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>361.772000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2004</th>\n",
       "      <td>1992.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>430.758000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2005</th>\n",
       "      <td>869.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>483.032000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"10\" valign=\"top\">Tajikistan</th>\n",
       "      <th>1996</th>\n",
       "      <td>1707.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>183.374000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1997</th>\n",
       "      <td>61.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>192.580000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1998</th>\n",
       "      <td>357.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>222.282000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1999</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>177.712000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000</th>\n",
       "      <td>46.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>159.417000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2001</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>167.035000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2002</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>187.958000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2003</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>236.394000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2004</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>308.844000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2005</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>337.129000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"10\" valign=\"top\">Syria</th>\n",
       "      <th>2011</th>\n",
       "      <td>4145.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012</th>\n",
       "      <td>51061.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2013</th>\n",
       "      <td>75145.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2014</th>\n",
       "      <td>75994.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2015</th>\n",
       "      <td>56738.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2016</th>\n",
       "      <td>50520.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2017</th>\n",
       "      <td>36442.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018</th>\n",
       "      <td>20235.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2019</th>\n",
       "      <td>10460.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2020</th>\n",
       "      <td>5624.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1396.944261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Sudan</th>\n",
       "      <th>2023</th>\n",
       "      <td>3463.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>536.922000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">Sri Lanka</th>\n",
       "      <th>2006</th>\n",
       "      <td>2146.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1698.156000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2007</th>\n",
       "      <td>2542.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1928.714000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2008</th>\n",
       "      <td>8388.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2410.801000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2009</th>\n",
       "      <td>10228.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2472.242000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2010</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2894.260000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2011</th>\n",
       "      <td>2.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3320.074000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3446.355000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 total_deaths  conflict_dummy conflict_onset          gdp\n",
       "country    year                                                          \n",
       "Ukraine    2022       73820.0               1              1  4582.609000\n",
       "           2023       70809.0               1              0  5337.491000\n",
       "Uganda     1996        1743.0               1              1   401.413000\n",
       "           1997        1313.0               1              0   423.918000\n",
       "           1998        1202.0               1              0   402.455000\n",
       "           1999         820.0               0              0   370.515000\n",
       "           2000        1157.0               1              0   355.466000\n",
       "           2001         212.0               0              0   347.812000\n",
       "           2002        1309.0               1              0   361.953000\n",
       "           2003        1592.0               1              0   361.772000\n",
       "           2004        1992.0               1              0   430.758000\n",
       "           2005         869.0               0              0   483.032000\n",
       "Tajikistan 1996        1707.0               1              1   183.374000\n",
       "           1997          61.0               0              0   192.580000\n",
       "           1998         357.0               0              0   222.282000\n",
       "           1999           1.0               0              0   177.712000\n",
       "           2000          46.0               0              0   159.417000\n",
       "           2001           0.0               0              0   167.035000\n",
       "           2002           0.0               0              0   187.958000\n",
       "           2003           0.0               0              0   236.394000\n",
       "           2004           0.0               0              0   308.844000\n",
       "           2005           0.0               0              0   337.129000\n",
       "Syria      2011        4145.0               1              1  1396.944261\n",
       "           2012       51061.0               1              0  1396.944261\n",
       "           2013       75145.0               1              0  1396.944261\n",
       "           2014       75994.0               1              0  1396.944261\n",
       "           2015       56738.0               1              0  1396.944261\n",
       "           2016       50520.0               1              0  1396.944261\n",
       "           2017       36442.0               1              0  1396.944261\n",
       "           2018       20235.0               1              0  1396.944261\n",
       "           2019       10460.0               1              0  1396.944261\n",
       "           2020        5624.0               1              0  1396.944261\n",
       "Sudan      2023        3463.0               1              1   536.922000\n",
       "Sri Lanka  2006        2146.0               1              1  1698.156000\n",
       "           2007        2542.0               1              0  1928.714000\n",
       "           2008        8388.0               1              0  2410.801000\n",
       "           2009       10228.0               1              0  2472.242000\n",
       "           2010           0.0               0              0  2894.260000\n",
       "           2011           2.0               0              0  3320.074000\n",
       "           2012           0.0               0              0  3446.355000"
      ]
     },
     "execution_count": 278,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "joined_df.head(40)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Next, the regression model (local projection) is specified"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The regression formula should look similar to this:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "$$\n",
    "y_{it+h} - y_{it-1} = \\beta_1^h c_{it} + \\beta_2^h c_{it-1} + \\sum_{j=1}^{h-1} \\beta_j^h c_{it+h-j} + \\theta_1^h \\Delta y_{it-1} + \\theta_2^h x_{it} + \\mu_i^h + \\gamma_t^h + \\epsilon_{it}^h\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [
    {
     "ename": "MissingDataError",
     "evalue": "exog contains inf or nans",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mMissingDataError\u001b[0m                          Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[276], line 12\u001b[0m\n\u001b[1;32m     10\u001b[0m x \u001b[38;5;241m=\u001b[39m pd\u001b[38;5;241m.\u001b[39mDataFrame({\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mc\u001b[39m\u001b[38;5;124m\"\u001b[39m: c, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mc_shift\u001b[39m\u001b[38;5;124m\"\u001b[39m: c_shift, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mdelta_y\u001b[39m\u001b[38;5;124m\"\u001b[39m: delta_y, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124msum_term\u001b[39m\u001b[38;5;124m\"\u001b[39m: sum_term})\n\u001b[1;32m     11\u001b[0m x\u001b[38;5;241m.\u001b[39msort_index(ascending\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m, inplace\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\n\u001b[0;32m---> 12\u001b[0m mod \u001b[38;5;241m=\u001b[39m \u001b[43msm\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mOLS\u001b[49m\u001b[43m(\u001b[49m\u001b[43my\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mx\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mentity_effects\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtime_effects\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m)\u001b[49m\n\u001b[1;32m     13\u001b[0m res \u001b[38;5;241m=\u001b[39m mod\u001b[38;5;241m.\u001b[39mfit()\n\u001b[1;32m     14\u001b[0m \u001b[38;5;28mprint\u001b[39m(\u001b[38;5;124mf\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mResults for h = \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mh\u001b[38;5;132;01m}\u001b[39;00m\u001b[38;5;124m\"\u001b[39m)\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/regression/linear_model.py:923\u001b[0m, in \u001b[0;36mOLS.__init__\u001b[0;34m(self, endog, exog, missing, hasconst, **kwargs)\u001b[0m\n\u001b[1;32m    920\u001b[0m     msg \u001b[38;5;241m=\u001b[39m (\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mWeights are not supported in OLS and will be ignored\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m    921\u001b[0m            \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mAn exception will be raised in the next version.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[1;32m    922\u001b[0m     warnings\u001b[38;5;241m.\u001b[39mwarn(msg, ValueWarning)\n\u001b[0;32m--> 923\u001b[0m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mOLS\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmissing\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmissing\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    924\u001b[0m \u001b[43m                          \u001b[49m\u001b[43mhasconst\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mhasconst\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    925\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mweights\u001b[39m\u001b[38;5;124m\"\u001b[39m \u001b[38;5;129;01min\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_init_keys:\n\u001b[1;32m    926\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_init_keys\u001b[38;5;241m.\u001b[39mremove(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mweights\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/regression/linear_model.py:748\u001b[0m, in \u001b[0;36mWLS.__init__\u001b[0;34m(self, endog, exog, weights, missing, hasconst, **kwargs)\u001b[0m\n\u001b[1;32m    746\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    747\u001b[0m     weights \u001b[38;5;241m=\u001b[39m weights\u001b[38;5;241m.\u001b[39msqueeze()\n\u001b[0;32m--> 748\u001b[0m \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mWLS\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmissing\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmissing\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    749\u001b[0m \u001b[43m                          \u001b[49m\u001b[43mweights\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mweights\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mhasconst\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mhasconst\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    750\u001b[0m nobs \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexog\u001b[38;5;241m.\u001b[39mshape[\u001b[38;5;241m0\u001b[39m]\n\u001b[1;32m    751\u001b[0m weights \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mweights\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/regression/linear_model.py:202\u001b[0m, in \u001b[0;36mRegressionModel.__init__\u001b[0;34m(self, endog, exog, **kwargs)\u001b[0m\n\u001b[1;32m    201\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, endog, exog, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m--> 202\u001b[0m     \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mRegressionModel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    203\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpinv_wexog: Float64Array \u001b[38;5;241m|\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m    204\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_data_attr\u001b[38;5;241m.\u001b[39mextend([\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mpinv_wexog\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mwendog\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mwexog\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mweights\u001b[39m\u001b[38;5;124m'\u001b[39m])\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/model.py:270\u001b[0m, in \u001b[0;36mLikelihoodModel.__init__\u001b[0;34m(self, endog, exog, **kwargs)\u001b[0m\n\u001b[1;32m    269\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m__init__\u001b[39m(\u001b[38;5;28mself\u001b[39m, endog, exog\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mNone\u001b[39;00m, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m--> 270\u001b[0m     \u001b[38;5;28;43msuper\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[38;5;21;43m__init__\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    271\u001b[0m     \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39minitialize()\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/model.py:95\u001b[0m, in \u001b[0;36mModel.__init__\u001b[0;34m(self, endog, exog, **kwargs)\u001b[0m\n\u001b[1;32m     93\u001b[0m missing \u001b[38;5;241m=\u001b[39m kwargs\u001b[38;5;241m.\u001b[39mpop(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mmissing\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mnone\u001b[39m\u001b[38;5;124m'\u001b[39m)\n\u001b[1;32m     94\u001b[0m hasconst \u001b[38;5;241m=\u001b[39m kwargs\u001b[38;5;241m.\u001b[39mpop(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mhasconst\u001b[39m\u001b[38;5;124m'\u001b[39m, \u001b[38;5;28;01mNone\u001b[39;00m)\n\u001b[0;32m---> 95\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdata \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_handle_data\u001b[49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmissing\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mhasconst\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     96\u001b[0m \u001b[43m                              \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     97\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mk_constant \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdata\u001b[38;5;241m.\u001b[39mk_constant\n\u001b[1;32m     98\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexog \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mdata\u001b[38;5;241m.\u001b[39mexog\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/model.py:135\u001b[0m, in \u001b[0;36mModel._handle_data\u001b[0;34m(self, endog, exog, missing, hasconst, **kwargs)\u001b[0m\n\u001b[1;32m    134\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m_handle_data\u001b[39m(\u001b[38;5;28mself\u001b[39m, endog, exog, missing, hasconst, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs):\n\u001b[0;32m--> 135\u001b[0m     data \u001b[38;5;241m=\u001b[39m \u001b[43mhandle_data\u001b[49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmissing\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mhasconst\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    136\u001b[0m     \u001b[38;5;66;03m# kwargs arrays could have changed, easier to just attach here\u001b[39;00m\n\u001b[1;32m    137\u001b[0m     \u001b[38;5;28;01mfor\u001b[39;00m key \u001b[38;5;129;01min\u001b[39;00m kwargs:\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/data.py:675\u001b[0m, in \u001b[0;36mhandle_data\u001b[0;34m(endog, exog, missing, hasconst, **kwargs)\u001b[0m\n\u001b[1;32m    672\u001b[0m     exog \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39masarray(exog)\n\u001b[1;32m    674\u001b[0m klass \u001b[38;5;241m=\u001b[39m handle_data_class_factory(endog, exog)\n\u001b[0;32m--> 675\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mklass\u001b[49m\u001b[43m(\u001b[49m\u001b[43mendog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mexog\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mexog\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmissing\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmissing\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mhasconst\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mhasconst\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m    676\u001b[0m \u001b[43m             \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/data.py:88\u001b[0m, in \u001b[0;36mModelData.__init__\u001b[0;34m(self, endog, exog, missing, hasconst, **kwargs)\u001b[0m\n\u001b[1;32m     86\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mconst_idx \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m     87\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mk_constant \u001b[38;5;241m=\u001b[39m \u001b[38;5;241m0\u001b[39m\n\u001b[0;32m---> 88\u001b[0m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_handle_constant\u001b[49m\u001b[43m(\u001b[49m\u001b[43mhasconst\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     89\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_check_integrity()\n\u001b[1;32m     90\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_cache \u001b[38;5;241m=\u001b[39m {}\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/statsmodels/base/data.py:134\u001b[0m, in \u001b[0;36mModelData._handle_constant\u001b[0;34m(self, hasconst)\u001b[0m\n\u001b[1;32m    132\u001b[0m exog_max \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39mmax(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexog, axis\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m0\u001b[39m)\n\u001b[1;32m    133\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m np\u001b[38;5;241m.\u001b[39misfinite(exog_max)\u001b[38;5;241m.\u001b[39mall():\n\u001b[0;32m--> 134\u001b[0m     \u001b[38;5;28;01mraise\u001b[39;00m MissingDataError(\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mexog contains inf or nans\u001b[39m\u001b[38;5;124m'\u001b[39m)\n\u001b[1;32m    135\u001b[0m exog_min \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39mmin(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mexog, axis\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m0\u001b[39m)\n\u001b[1;32m    136\u001b[0m const_idx \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39mwhere(exog_max \u001b[38;5;241m==\u001b[39m exog_min)[\u001b[38;5;241m0\u001b[39m]\u001b[38;5;241m.\u001b[39msqueeze()\n",
      "\u001b[0;31mMissingDataError\u001b[0m: exog contains inf or nans"
     ]
    }
   ],
   "source": [
    "# now, the local projections formula is specified\n",
    "\n",
    "for h in range(1, 11):\n",
    "    c = joined_df[\"conflict_dummy\"]\n",
    "    c_shift = joined_df[\"conflict_dummy\"].shift(1)\n",
    "    delta_y = np.log(joined_df[\"gdp\"].shift(1)) - np.log(joined_df[\"gdp\"].shift(2))\n",
    "    sum_term = sum([joined_df[\"conflict_dummy\"].shift(h-j) for j in range(1, h)])\n",
    "    y = np.log(joined_df[\"gdp\"].shift(-h)) - np.log(joined_df[\"gdp\"].shift(1))\n",
    "\n",
    "    x = pd.DataFrame({\"c\": c, \"c_shift\": c_shift, \"delta_y\": delta_y, \"sum_term\": sum_term})\n",
    "    x.sort_index(ascending=False, inplace=True)\n",
    "    mod = sm.OLS(y, x, entity_effects=True, time_effects=True)\n",
    "    res = mod.fit()\n",
    "    print(f\"Results for h = {h}\")\n",
    "    print(res.summary)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}