The black line shows the unweighted average share of Communist votes across all communities. e blue and red lines show how communities with high levels of emigration to the West and communities with high levels of emigration to the East deviate from the overall trend. We plot residual shares of Communist votes controlling for the same set of pre-migration community-level variables as our baseline speci cation (see column 3 of Table A5) apart from the 1998 election results. Communities with high levels emigration to the West (East) are de ned as having an above median prevalence of westward (eastward) migration and above 50 percent share of westward (eastward) migrants among all migrants. Bars show the overall number of emigrants in stocks (in 1000). Data come from yearly waves of the Moldovan Labor Force Survey. Pre-2006 numbers of emigrants are adjusted to account for a change in the sampling method of the Moldovan Labor Force Survey. Data on emigration from Moldova before 1999 are not available. e rst wave of the Moldovan Labor Force Survey was conducted in 1999, just a er the unexpected Russian nancial crisis hit Moldova in late 1998 and triggered the rst big wave of emigration. Information on destination countries is not available in pre-2006 waves. e same trend in the number of migrants is observable using data on Moldova immigrants from major destination countries. In 1998, for example, only 15 Moldovan immigrants were registered in Italy. is number increased to 40,000 by 2004. A similar development occurred in other destination countries such as Greece, Portugal and Spain (see footnote 10 for sources and more details). e orange line shows the volume of international calls to Moldova (in 1000 hours per week). Data come the International Tra c Database compiled by Telegeography.
рыжее — просто соединение первой и последней точки