Does It Matter How Much The Government Helps Agriculture?
Does It Matter How Much The Government Helps Agriculture?
Governments in rich countries like to support farmers. Since the mid 1980s, the US government has sent about $15b a year directly to farmers (about 15% of average net farm income). Payments were unusually high in 2020 and 2021 due to COVID compensation and ad hoc disaster payments (the "ad hoc disaster" in this case was the trade war with China).
Does such government support change what farmers do, or is it just a nice perk? Last month in Agricultural Economics published a paper in which Nathan Hendricks, Nelson Villoria, Matthieu Stigler and I present some answers to this question.
Beginning with the 1996 Freedom to Farm Act, payments to US farmers became decoupled from current production (commodity programs) and were instead tied to historical production (base acre programs). This change decoupled current production decisions from government payments and, in theory, enabled fair global trade. In contrast, the pre-1996 strategy of paying farmers to produce certain commodities caused them to produce more of those commodities, which lowered the price and harmed producers in other countries.
There are two ways to increase production in response to government or other incentives: (i) use more inputs like land and fertilizer, or (ii) get more out of your inputs (improve productivity). Productivity increases are particularly valuable because they enable us to get more for less both now and in the future.
It works the other way too. Governments that suppress the returns to agriculture by, for example, taxing exports may reduce the incentive for productivity improvements. In his 1980 Nobel Prize lecture, T.W. Schultz lamented how governments in low-income countries dissuaded productivity-enhancing investments.
“For want of profitable incentives, farmers have not made the necessary investments, including the purchase of superior inputs. Interventions by governments are currently the major cause of the lack of optimum economic incentives.” – T.W. Schultz (Nobel Prize Lecture)
That's the theory, but what happens in the real world?
The graphs below show some measures of government support for agriculture in Europe, North America, and Oceania (rich countries). The graphs show the Nominal Assistance Coefficient (NAC), which equals the price that producers in a country receive relative to the world price that would exist under free trade. If NAC=1, it means that the government does not distort farm prices; farmers receive whatever they would get on the world market.
All these rich countries have NAC>1, which means that government intervention raised the prices received by their farmers, often through production or export subsidies. Many of these governments, including the US, have reduced their rate of assistance recently as a part of free trade negotiations.
Rich country governments tend to give higher assistance rates to farmers who compete with importers in the domestic market than to farmers who tend to compete in the export market. (Specifically, NAC is higher for positive net import commodities than positive net export commodities.)
Countries in Sub-Saharan African tended to set NAC<1 by, for example, restricting exports. Such policies helped keep prices low for consumers. In the 1980s, the World Bank and International Monetary Fund began to make loans to these countries conditional on them adopting more market-oriented policies. As a result, NAC increased towards one in many of these countries, i.e., relative farm prices increased.
In our study, we want to know how much an increase in NAC causes input use, productivity, and production to increase.
Our challenge is the we often don't know why the NAC changed, which can be a big problem if we want to claim causality. For example, what if a country experiences a bad crop year due to a drought and the government increases farm assistance to compensate farmers for their market losses. We see NAC go up and production go down, but it was production causing NAC and not the opposite. To get around this problem, smooth out the annual fluctuations. We use a moving average of NAC in our analysis rather than the annual NAC (the solid rather than the dashed lines in the graphs).
Here's another example. What if a country gets richer by moving from agriculture to manufacturing, as many countries in Asia did during our sample period. Politicians may respond by increasing support for agriculture to preserve the little they have left. High price supports for rice in Japan, South Korea and Taiwan are an example of this phenomenon. In the data, we would see NAC decreasing and production increasing, but it would be production causing NAC not the other way around.
There's no easy way around this problem. We address it by adding control variables and fixed effects to our regression models, and by leaning on narratives about the causes of changes in NAC. For example, we are more confident inferring causality from Sub-Saharan Africa where large NAC changes were driven by World Bank and IMF directives than from Asia where NAC changes were mostly caused by production changes.
What do we find?
Reducing the anti-agricultural bias in Sub-Saharan Africa (i.e., increasing NAC towards 1) raised production, mostly by increasing productivity. A country in this region that increased the effective price by 10% for exportables would have an increase in the rate of productivity growth of 0.328 percentage points, which is especially large compared to the average historical growth rate of only 0.5% for this region.
In Asia, increasing NAC for importables is associated with a decrease in production. This likely reflects production declines causing NAC to increase, as explained above. In Europe and the West, most countries decreased their NAC over time. For importables, these decreases coincided with significant decreases in production growth. A little more than half of the decrease in production came from reduced input growth. There is not enough data in Latin America to obtain precise estimates.
So, does it matter how much governments support agriculture? Does an anti-agriculture bias suppress agricultural productivity? Our answer is yes and yes. In particular, productivity increases in Sub-Saharan Africa provide direct evidence in support of the Schultz thesis that removing disruptive government interventions improves productivity.
Citation: Hendricks, N.P., A. Smith, N.B. Villoria, and M. Stigler, "The effects of agricultural policy on supply and productivity: Evidence from differential changes in distortions" Agricultural Economics. 2022. https://doi.org/10.1111/agec.12741
How is NAC measured? We measure distortionary government payments using a unique dataset constructed by Kym Anderson and colleagues at the World Bank for 82 countries over the period 1961–2011. They construct a statistic called the Nominal Assistance Coefficient (NAC), which equals the price that producers in a country receive relative to the world price that would exist under free trade. The NAC captures distortions due to a wide range of government interventions including trade distortions, domestic subsidies or taxes, distortions to exchange rates, and distortions to the price of inputs.
Regression Models. We regress the change in the log outcome variable (production, input use, or total factor productivity) on the log of NAC for exportables and importables, country and region-year fixed effects, and controls including a polity score and quadratic functions of precipitation and temperature. The graph above shows coefficients from models estimated separately by region. See the paper for results from alternative specifications.
Funding: This work was supported by the USDA National Institute of Food and Agriculture, Agricultural and Food Research Initiative Competitive Program, Agriculture Economics and Rural Communities, grant # 2016-67023-24637.
Code: I made the US government support graph using this R code: govtsupport.R