Keywords

The main results of the Semi-Input-Output (SIO) multiplier analysis are displayed in Table 7.1. The table contains the results for three scenarios for all three years: (1) agriculture-led growth (a one-unit demand injection into agriculture), (2) manufacturing-led growth (a one-unit demand injection into manufacturing) with an unconstrained agricultural sector, and (3) manufacturing-led growth with a supply-constrained agricultural sector.

Table 7.1 SIO multiplier analysis
Full size table

Three aggregates are calculated based on the SIO model: total gross output multiplier, total GDP multiplier, and households’ total income multiplier. The output multiplier is the sum of the total effects in the activity accounts and shows how much total output would increase following a one-unit increase in demand for agriculture or manufacturing. Taking Column 1 as an example, the table shows that if one million birr had been injected into the agricultural sector in 2002 (Column 1), the result would have been an increase of 1.65 million birr in agriculture itself, 0.10 million birr in food-processing, 0.49 million birr in the trade and transport sector, 0.26 million birr in the private services sector, and smaller increases in the remaining sectors. Together, an increase of one million birr in exogenous demand for agriculture would have increased the gross output of the entire economy by 2.64 million birr. The one-unit demand increase is seen in the corresponding commodity demand cell for agriculture or manufacturing in row 11/13. Similar to the output multiplier, the GDP multiplier is the sum of the total effects in the factor payment accounts. The multiplier shows how much the income of production factors (labor and capital) would increase per one-unit demand increase and expresses the increase in the total value added. The households’ income multiplier is the sum of total effects in the household accounts, showing the total effect on urban and rural households’ income of a one-unit increase in demand. The GDP and income multipliers are smaller than the output multipliers due to different leakages in the circular flow of income (e.g., import and tax leakages), which is standard in most economic structures (Breisinger et al., 2009).

The results of the SIO multiplier analysis show that an exogenous injection of one million birr into agriculture in 2002 would have, given the linkage structure in Ethiopia that year, led to a total GDP increase of 1.99 million birr and an increase in total output of 2.64 million birr (Column 1). It would also have generated additional household incomes of 1.98 million (1.27 in urban areas and 0.71 in rural areas) after taking the various transfers, spillovers, and feedback effects in the economic system into account. The same injection of exogenous demand into the manufacturing sector in 2002 (Column 2) would only have led to an increase of 0.85 million birr in Ethiopia’s GDP, 1.42 million in increased output, and 0.85 million in household income (0.29 million in rural areas and 0.55 million in urban areas).

The higher multiplier effect of agriculture compared to manufacturing continues throughout the period of study. In both 2006 and 2010, the higher multiplier for agricultural demand compared to manufacturing demand remains. One million birr of increased exogenous demand for agricultural products would have led to an increase in GDP of 2.22 million birr in 2006 and 2.46 in 2010 (Columns 4 and 7). In contrast, increased manufacturing demand would have led to GDP increases of 0.58 million birr in 2006 and 0.63 million birr in 2010 (Columns 5 and 8). The output and income multipliers also remained smaller for manufacturing. If agriculture were a supply-constrained sector (so that increased demand is satisfied with exports), a manufacturing-led investment strategy would have even more limited linkages in Ethiopia during this time period. The total GDP multipliers in this scenario are 0.50 in 2002, 0.29 in 2006, and 0.29 in 2010 (Columns 3, 6, and 9).

These larger economic linkages for the agricultural sector compared to those of the manufacturing sector are linked to the structure of the Ethiopian economy. An analysis of the underlying Social Accounting Matrices (SAMs) reveals that three aspects are important features of this structure. First, agricultural products make up a large share of households’ spending in Ethiopia, which is consistent with the structure of many low-income countries. In Ethiopia, agricultural and food-processing products are the largest share of household consumption for both urban and rural households. On average, in the 2002–2010 period, this accounted for 60% of rural households’ spending and 45% of urban households’ spending (Table 7.2). Second, the SAMs reveal that agriculture is the most labor-reliant sector in the Ethiopian economy (Table 7.3); such sectors tend to be more linked to the domestic economy than capital. Third, a larger share of the increased demand translates into output increases in the agricultural sector than in the manufacturing sector (comparing the demand and output multipliers in Table 7.1). This implies that a larger share of increased investments into manufacturing would be leaked to imports rather than stimulating the domestic economy compared to increased investment in agriculture. Together, these structural features and the rural and agriculture-dominant character of the Ethiopian economy contribute to the larger multipliers for agriculture compared to manufacturing.

Table 7.2 Average share of household consumption spending by sector, 2002–2010 (%)
Full size table
Table 7.3 Average contribution of capital and labor to total value added by sector, 2002–2010 (%)
Full size table

Looking at the change over time, the multiplier analysis shows that in each of the three snapshots in 2002, 2006, and 2010, growth in the agricultural sector had a stronger effect on overall growth than growth in the manufacturing sector would have had. Rather than decreasing—as expected if the dynamism of the Ethiopian growth came from a sector outside agriculture—the agricultural multipliers increased. Over the same time period, the multipliers for manufacturing decreased. These results imply that in each of the three years, the agricultural sector was a better engine of overall growth than the manufacturing sector would have been. They also imply that during this period, the strength of the agricultural sector as a growth engine was not outpaced by the manufacturing sector.

While the results of the SIO multiplier analysis highlight the importance of the agricultural sector as a growth engine, they also shed light on a possible challenge for Ethiopia’s continued economic growth. For growth linkages to stimulate growth—whether stemming from the agricultural sector or otherwise—it is crucial that the sectors that these linkages stimulate can grow; otherwise, the linkages will not result in aggregate growth. The decreasing growth linkages for the manufacturing sector during this period might therefore be a cause of concern. If the economic growth were leading to a successful structural transformation toward higher-productivity sectors (such as manufacturing), one would expect the growth linkages of the manufacturing sector to have grown during this period.

Discussion of Results

The results of the SIO multiplier model show that the agricultural sector was the best growth option in 2002, 2006, and 2010, with high and strengthening growth linkages to other sectors. Due to the apparent rigidity of the manufacturing sector, manufacturing growth would not have had as strong an effect on overall growth.

The study’s findings are largely in line with the substantial body of previous work suggesting that agriculture has large multipliers in low-income countries (Pyatt & Round, 1979; Hazell & Roell, 1983; Haggblade et al., 1991; Powell & Round, 2000; Diao et al., 2010b). The findings are also in line with Diao et al.’s (2007, 2010a) findings that agriculture-led growth has been broadly successful in generating growth in Ethiopia given its economic structure. Moreover, the present work provides a more formal SAM-based model for Dorosh and Mellor’s (2013) finding that agriculture is a viable means for growth in Ethiopia.

In addition to the inherent limitations of the SIO methodology discussed in Chap. 6, the study’s application of the chosen method could have been strengthened by a finer disaggregation of regions, production for market and home consumption, sub-sectors, and by extending the timespan covered. However, such disaggregation was not possible given the available SAMs and the need for a similar SAM structure over time. Despite these limitations, the empirical investigation is based on rich data for the three years under study, and offers an extension of the previous one-year studies exploring agriculture’s growth linkages in Ethiopia. The changes that the Ethiopian agricultural sector underwent during the study period—with substantial growth in agricultural production, productivity, and input use; changing demand patterns; and growing urbanization (Tamru et al., 2017; Bachewe et al., 2018; Dorosh et al., 2018; Vandercasteelen et al., 2018)—lend further support to that the increased agricultural growth linkages that the study identifies are plausible. As such, despite limitations, the applied methodology provides several insights into the functioning of the Ethiopian economy.

The findings show that the agricultural sector has been an important engine of Ethiopia’s growth in the short- to medium-term perspective that is studied. However, it cannot speak to whether this also holds for the long term; the time frame is too short, and SIO would not be the optimal method for long-term analysis. The results that the study does provide suggest that while agricultural growth played a large role in Ethiopia’s economic growth in 2002–2010, this agriculture-led growth did not spur a structural transformation away from agriculture, as the growth linkages from agriculture increased while those of the manufacturing sector decreased. Based on theory and the historical experience of most now-rich countries, successful long-term growth requires a structural transformation away from low-productivity agriculture to more productive sectors. Historically, this has meant a structural shift toward the manufacturing sector (Chenery & Syrquin, 1975; Rodrik, 2013). However, while some scholars still advocate for the primacy of manufacturing-led growth (Lin, 2012, 2015), it is unclear whether this pattern will hold in the future. The low-income countries of today have seen a much more limited experience in the manufacturing sector than previous developers (Rodrik, 2016; Gollin et al., 2016). Going forward, sectors such as the service sector or high-productive agriculture may be able to take on some of the beneficial characteristics historically associated with manufacturing (Gollin, 2018). If sufficiently permissive conditions are created for the service and agricultural sectors, there is no inherent reason why they cannot be important ladders to economic growth in today’s low-income countries. Given the rapid development of the agricultural sector in Ethiopia in the last 20 years (Bachewe et al., 2018; Rohne Till, 2021) and the limited size of the manufacturing sector in terms of employment and output (Table 5.2), growth linkages (Table 7.1), and low creation of formal employment opportunities (Diao et al., 2021), this ought to be good news for Ethiopia’s future growth prospects.

However, even if these aspects indicate that the agricultural sector could continue to be an important engine of growth beyond the short and medium term studied, this prospect does not come without challenges. In addition to the lack of historical precedence—no now-rich country has achieved this status based on agricultural growth in the long term—two aspects are of key concern. First, if agricultural growth continues without growth in other sectors, the falling relative prices of agricultural products may undermine the ability of agricultural growth to lead to overall growth. In order to sustain overall growth, nonagricultural growth is also needed to match the growing supply of agricultural products and increasing demand for nonagricultural products as a result of agricultural growth. Second, the state of Ethiopia’s infrastructure might limit the extent to which agricultural growth can continue to generate aggregate growth. Virtually all previous research on agricultural growth linkages emphasizes the importance of rural infrastructure (Haggblade et al., 2007). Despite recent improvements and substantial public spending, Ethiopia still has one of the lowest road densities in the world and has high transport costs relative to international standards (Minten et al., 2014). Lacking infrastructure limits market connections and leads to poorly functioning commodity and factor markets, which limit the potential for agricultural growth to successfully translate into growth in nonagricultural sectors.

These results suggest that while agriculture has been the main sector of growth in the medium term at an early stage of economic development in Ethiopia, it likely cannot be the sole engine of growth for successful long-term economic growth and structural transformation. The realization of agriculture-led aggregate growth will depend on growth both inside and outside of agriculture. This implication is also in line with the previous literature on agriculture-led growth in Ethiopia; for successful growth, agriculture cannot be focused on in isolation of the rest of the economy (Dercon et al., 2009; Diao et al., 2010a). The relatively poor performance of the Ethiopian manufacturing sector so far warrants further investigation. Under a successful process of growth and structural transformation, this sector would also thrive. Its lack of success may itself reflect the challenges identified above, which may have limited the ability of agricultural growth to translate into growth in nonagricultural sectors.

In light of the empirical results and the discussion of these, the main conclusion to emerge from the empirical analysis is that the agricultural sector was the best growth engine in Ethiopia in the studied period. In 2002–2010, the agricultural sector had high growth linkages, which did not diminish during the growth process that took place during these years.