Columns VIII-IX introduce government consumption and total fertility; again the results match those of Table 7, though the coefficient on yield remains consistent but is no longer significant at the 10 percent level. Finally, just as in Table 7, Columns X-XI drop the government consumption variable and report a coefficient of 0.35, now significant at the 10 percent level and consistent in magnitude with Table 7. Overall, the results using a 10-year lag on yield remain highly consistent with the results in Table 7, though the statistical threshold for significance is not passed in two of the second stage specifications. Finally, Table 10 presents a NAVA growth framework using GMM instrumentation and finds similar agricultural productivity effects on value added in non-agriculture sectors. Column I runs difference GMM and finds that a 10-year lag on yield is associated with subsequent increases in non-agricultural value added per worker, significant just short of 5 percent levels. The coefficient of 0.1 suggests that a 0.5 ton increase in yields leads to a 5 percent higher non-agricultural labor productivity 10 years later, which translates to a 0.5 percentage point higher growth rate. Note that this magnitude lies between the fixed effects coefficients of 0.05-0.06 and the IV coefficients of 0.27- 0.37 in Table 6, adding support to the overall results. The specification in Column I passes the Sargan test for over identification of instruments with a p-value of 0.43. Column II employs the Blundell-Bond “system” GMM estimator, though this does not pass a Sargan test under any relevant specification, so we prefer to interpret only difference GMM specifications. Column III adds the fertilizer price instrument to the exogenous variables in the specification, and finds similar results to Column I. Again, the estimation passes a Sargan test, and the AR test is satisfied with a pvalue of 0.09. Our analysis documents the strong positive links between agronomic inputs—fertilizer, water and modern seeds—and cereal yields per hectare, even after a variety of controls are introduced. We employ a combination of fixed effect,gutter berries instrumental variable and Arellano-Bond GMM estimators to posit a causal economy-wide link between, first, input use and yields, and, second, yields and various measures of economic growth and structural change.

We construct a novel instrument exploiting the economic geography of fertilizer production, which together with global fertilizer price fluctuations allow us to study economic growth and structural change in a statistically causal framework. The cross-country substantiation of both agricultural yield production functions and their links to various dimensions of economic growth and structural change are novel empirically. Taking the coefficients from Table 4, a representative country with yields of 1 t/ha that introduces an input package to jump from, say, 15 kg/ha to 65 kg/ha of fertilizer use would be expected to see an average yield jump of 147-470 kg/ha; while increasing from 10 to 50 percent use of modern seed would be expected to increase yields by 480 kg/ha. On the economic growth side, the instrumental variable results suggest that boosting yields from 1.5 t/ha to 2.0 t/ha is linked to a range of 13 to 19 percent increase in income per capita, a 3.3 to 3.9 percentage point lower share of labor in agriculture five years later, and approximately 20 percent higher non-agricultural labor productivity after roughly one decade. The estimated effects are identified based on exogenous variation in fertilizer prices, and are robust to the inclusion of controls for investment and standard macroeconomic policy indicator variables. The results suggest that land productivity promotes growth both by supporting changing labor shares and by increasing total factor productivity. Regressions focused on marginal effects of individual variables are, of course, not intended to evaluate nonlinear outcomes guided by Leontief-style agricultural production functions and discontinuous policy functions, so the regression results might underestimate the potential effects of yields. The results might also be constrained by issues of heterogeneity in cross-country production functions . The evidence in this paper points to strong potential yield and growth effects resulting from policy efforts to support adoption of a green revolution-type package of inputs in economies with low agricultural productivity and a large share of the labor force still in agriculture. The results suggest a particularly strong role for fertilizer, which is highly consistent with field station agronomic evidence.

Fertilizer’s high private return on experimental plots and in the field suggests some sort of market failure that policy can address; scholars debate whether the failure is due to credit constraints or non-rational behavior on the part of farmers . Regardless, the evidence presented in this paper suggests social returns from fertilizer use that exceed the immediate private returns, furthering the case for policy efforts. It is worth briefly describing the main concerns about increasing fertilizer use. One set is environmental. These are legitimate and require foresight in policy planning, but as Palm et al. have indicated, countries should not simply avoid fertilizers for environmental reasons, since soil degradation induced by fertilizer omission poses much a greater risk to agricultural production. A second class of concerns focuses on inequality and the potential scale bias of modern inputs. Hayami and Ruttan review the evidence on the alleged scale bias in the Asian green revolution and find that the evidence does not support this allegation. A third set of concerns focuses on both the challenges of governments implementing input support programs and also the challenges of exiting from them in due course. Though there is evidence that subsidy programs can be successful , there is also evidence that they can be subject to elite capture, and there is concern that their fiscal drag effects can far outlive their usefulness . While our results provide some evidence for a causal link from agricultural productivity increases to structural change and higher non-agricultural labor productivity, we can only speculate on the mechanisms through which these effects play out. Nevertheless, our novel identification of a causal link from yield increases to labor composition shift and non-agricultural productivity increases rules out models where structural change is driven solely by “pull” forces from growing non-agricultural sectors. To the extent that our results show that yield increases contribute to increases in non-agricultural labor productivity growth, this suggests that structural change involves more than just the satiation of food needs and the movement of labor into other sectors. This labor share shift somehow accelerates labor productivity growth. One possible mechanism might be increasing returns in the non-agricultural sector, perhaps through learning-by-doing as in the example modeled in Section 2 of this paper. Perhaps increased food production lowers average prices and frees up consumers’ resources for other consumption and for productive public and private investments, raising labor productivity elsewhere. Or perhaps higher availability of staple foods promotes improved health and labor productivity across sectors. Identifying more precise causal pathways between staple yields and structural change forms an important topic for future work.

Karadimou and Markatos developed a transient two-phase model to study particle distribution in the indoor environment using Large Eddy Simulation method. Baek et al. used CFD analysis to study various combinations of air conditioners and fans to improve growth rate in a plant factory. More recently, Niam et al. performed numerical investigation and determined the optimum position of air conditioners in a small vertical plant factory is over the top. In addition, a variety of mathematical techniques are proposed to provide sub-model for investigating photosynthesis. According to Boulard et al., tall canopies can induce a stronger cooling of the interior air by using a CFD model to study the water vapor, temperature, and CO2 distribution in a Venlo-type semi-closed glass greenhouse. Despite the fact that photosynthesis plays an integral role in distribution of species and uniformity along cultivation trays, this issue has not been well addressed. Although numerous research works have been done to investigate the turbulent flow in enclosures and buildings, this study is the first to numerically investigate the transport phenomena considering the product generation and reactant consumption through photosynthesis and plants transpiration with CFD simulations for IVFS-based studies. Furthermore, a newly proposed objective uniformity parameter is defined to quantify velocity uniformity for individual cultivation trays. Moreover, numerical simulations are performed to simulate and optimize fluid flow and heat transfer in an IVFS for eight distinct placements of flow inlets and outlets in this study. Accordingly, the effects of each case on uniformity, relative humidity, temperature,strawberry gutter system and carbon dioxide concentration are discussed in detail. Finally, an overall efficiency parameter is defined to provide a holistic comparison of all parameters and their uniformity of each case.In this study, three-dimensional modeling of conjugated fluid flow and heat transfer is performed to simulate the turbulent flow inside a culture room having four towers for hydroponic lettuce growth. Assuming that the four towers are symmetric, a quarter of the room with four cultivation trays is selected as the computational domain, as illustrated in Fig. 1a.The effect of LED lights on heat transfer is considered through constant heat flux boundary conditions at the bottom surface of each tray as shown in Fig. 1b. Lastly, the species transfer due to photosynthesis are occurring only in the exchange zone, which is illustrated in Fig. 1c. To study the impact of air inlet/exit locations on characteristics of air flow, four square areas, denoted as A, B, C, and D in Fig. 1a, are considered to be inlet, exit, or wall. To perform a systematic study, Table 1 presents the location of inlet and exit for all eight cases studied. With the aim of comparing all of the proposed designs, case AB is selected to be the baseline.To consider the effect of heat transfer with the outdoor ambient air, a solid wall zone comprised of plywood with a thickness of 0.12 m is assumed in the simulation. A constant-temperature boundary condition is set on the outer surface of the wall. Both conduction and convective heat transfer are considered within the model. All dimensions and boundary conditions are listed in Table 2.

In our model, the species exchange zone of photosynthesis is defined to be directly above the upper surface of each tray. These zones have the same cross-sectional area as the trays with the height of 0.1 m. Within the exchange zone, the water transpiration rate and carbon dioxide consumption rate are defined according to the experimental data obtained by Jin et al.and Adeyemi et al..One of the most critical factors affecting crop growth rate is the air flow velocity over plants. A fluid stream with horizontal speed ranging from 0.3 to 0.5 m s−1 can escalate the species exchange between the flow and plant leaves resulting in enhancement of photosynthesis. In indoor farming systems, the flow velocity can be controlled well using ventilation fans for more efficient plant growth. However, heterogeneous distribution of feeding air over plant trays can cause undesirable non-uniformity in crop production, which should be avoided. Therefore, it is important to study the effect of inlet-outlet location and flow rate on the flow patterns throughout the culture room. Herein, the most favorable condition is defined as the condition at which the flow velocity above all trays is equal to the optimum speed Uo, which is set to be 0.4 m s−1. The objective uniformity, OU, defined in Eq. is used to assess the overall flow conditions. The OU for all eight cases as a function of mass flow rate are summarized in Fig. 5. Since the inlet/exit area and air density remain the same, the mass flow rate is directly proportional to flow velocity. In addition, the target flow velocity over the plants is set to be 0.4 m s−1. Therefore, a general trend of OU first increases and then decreases when increasing the overall mass flow rate. Depending on the design, the peak of OU occurs at different mass flow rate for each case. Another general trend can be observed that the peak of OU occurs at a lower mass flow rate if the inlet is located at the top due to buoyancy force. This can be clearly demonstrated by cases AB and BA or AD and DA . Therefore, there exists a different optimal inlet/exit design for each mass flow rate condition. As can be seen from Fig. 5, the maximum OU at flow rates of 0.2, 0.3, 0.4 and 0.5 kg s−1 is observed for configurations AD, BC, BA, and DA, respectively. Therefore, this simulation model can identify optimal flow configuration at a specific mass flow rate condition.