Remittances, the part of the migrant's income sent back to their family living in the origin country, have become a critical stepping-stone to economic development for many developing nations. A key factor that causes migrants to use informal channels is the high cost of transferring funds through formal channels. Reducing the cost of remitting is one of the 2030 Sustainable Development Goals; it is also an important policy objective as it helps to bring remittances into the formal economy, enhances financial inclusion and increases the net income of receiving households. This study examines the question of whether and to what extent the reduction in the cost of remittances increases the flow of remittances to developing countries, and whether larger amounts are remitted when the cost per transaction decreases (the so-called scale effect). It uses bilateral data on remittance flows and exploits a novel dataset covering transaction costs for 30 sending and 75 receiving countries for the period 2011-2017. A gravity model of remittance flows is estimated using panel data and instrumental variable techniques to account for potential endogeneity. We find that transaction cost is a significant predictor of the volume of formal remittances. A 1 percent decrease in the cost of remitting USD 200 leads to about a 1.6 percent increase in remittances. This association remains unchanged regardless of the models used and techniques employed. In addition to this strong impact of transfer fees, migrant stock, exchange rate stability in the recipient country and financial development in both the recipient and sending countries are also found to be important factors driving remittances. The findings suggest that policies designed to increase remittances need to focus on decreasing the cost of remitting through formal channels.