What's good for countries is good for communities

A new paper by Trevor Tombe in the American Economic Journal: Macroeconomics provides a cautionary tale for those who think we'd be better off economically if we sourced more food closer to home.  

The paper starts off with this stylized fact

Agriculture in poor countries has low productivity and high employment relative
to other sectors. Differences in aggregate productivity and income between
rich and poor countries are therefore primarily due to differences within agriculture;
Schultz (1953) calls this the Food Problem

The problem, it seems, is that poor, agricultural-based countries don't trade.

Trade data, though, presents a puzzle: with low relative agricultural
productivity, developing countries should be massive food importers; yet, they
are not (see section I.B). The “missing” food imports suggests trade costs may
be important for international productivity differences.

These "trade costs" that are holding up trade include bad policies like tariffs but also bureaucratic problems like border delays.

The paper has a nice set graphs, and the following shows a few of them related specifically to agriculture.

First, when you don't trade agricultural products with you're neighbors, you tend to be poor.

Countries that have more agricultural trade have higher GDP/worker.  The correlation is 0.68.  Countries that have more agricultural trading partners tend to be richer and more productive.

A lot of local food advocacy is premised on the idea that we need smaller farms with more farmers.  I don't think anyone would argue that greater local food production would mean  more employment in agriculture.  Well, would that be good for the economy?  According to this paper, no. This graph shows a very clear negative correlation between a country's employment share in agriculture and GDP/worker.  More workers in agriculture is typically going to mean lower economic output.  

Finally, this graph shows that the higher the share of food spending on domestic output (i.e., the more local foods purchased), the lower the country's GDP/worker.   

There's no reason to believe what's true of countries wouldn't also be true for states or cities trading with each other.

The author concludes:

Agricultural trade costs account for roughly 25 percent of the aggregate differences between rich and poor countries. Trade costs in agriculture and manufacturing together account for over 40 percent. Even observable, policy-relevant trade costs (tariffs and border
delays) have important negative productivity effects in poor countries. The food
trade is missing in poor countries and in our models; it should be no longer.