Source: See fig. 4.

Notes: In panel A per capita GDP comes from WDI. Economy-wide capital labor ratios for Tanzania and Ethiopia are calculated from the IMF Investment and Capital Stock Dataset (2017) capital information and ETD data on national employment, and for the Czech Republic the EU KLEMS is used. In panel B all data for Tanzania and Ethiopia come from the firm-level census data, while the Czech Republic comes from EU KLEMS. In panel A, the values are relative to the Czech Republic, for which each variable = 100. For example, per capita GDP in Tanzania is 4.9 percent of per capita GDP in the Czech Republic. In panel B 2010–2017 average capital-labor (K/L) ratios are measured in $1,000 constant 2010 PPP per worker terms for Ethiopia and 2010–2016 for Tanzania; these values are then reported relative to the Czech Republic, for which each variable = 100; sector-specific levels are used for the Czech Republic in columns 2–4. Average capital-labor (K/L) ratio for a group of firms is an employment-weighted ratio using K/L for firms and their employment shares within that group. Firms with 10–49 employees are classified as small firms while firms with 50 or more employees are classified as large firms. The Tanzania employment numbers used in these calculations cover all workers on contracts longer than one month, while the Ethiopia employment numbers are limited to permanent workers; the analysis uses permanent workers to be consistent with the Czech Republic data from EU KLEMS. Foreign firms are defined as those with foreign ownership for the majority of years or at least one year with data available only for two years. Exporters are firms reporting exports in every year. The definitions used for foreign and exporting firms are constructed to create a time-invariant variable in order to track the same group of firms over time and produce consistent aggregate estimates. Among the large firms, the analysis further classifies firms by capital intensity in the following way. First, it ranks firms by capital-labor ratio (K/L) and then chooses the top 10 percent firms according to K/L for the initial year of each panel (1996 for Ethiopia and 2008 for Tanzania). In the next year, the analysis adds to these new firms according to their K/L ratio in the same way. Thus, once a firm is classified as being in the top 10 percent according to capital intensity, it remains in that group for every year it is observed. The same procedure is followed to classify the bottom 10 percent of firms, and the middle 80 percent is defined as the residual. Among the large firms, the analysis further classifies firms that enter the sample in 2010 or later as new entrants relative to firms that are in the sample prior to 2010.