The United States plants more than 170 mil­lion acres of corn and soy­beans a year, more than any coun­try in the world. And the pri­ma­ry mech­a­nism in the US that we use to sub­si­dize agri­cul­ture is actu­al­ly called the Federal Crop Insurance Program. So, the crop insur­ance pro­gram in the US is also the largest such pro­gram glob­al­ly, with over $100 bil­lion in lia­bil­i­ties annu­al­ly. So it’s a very big program.

The way that crop insur­ance works, basi­cal­ly, is it insures farm yields and rev­enues from loss due to say, bad weath­er or low prices. It’s a public/private part­ner­ship, which works where the gov­ern­ment basi­cal­ly sets the pre­mi­um rates, pri­vate com­pa­nies deliv­er it, and then farm­ers pay part of the premium. 

So, the pri­ma­ry goals, his­tor­i­cal­ly, have been to improve, increase max­i­miza­tion of par­tic­i­pa­tion in the pro­gram, but less focus has been on the mod­ern­iza­tion of the rat­ing sys­tem. So I’ll just give you an exam­ple of how it’s sort of crude in some ways.

The gov­ern­ment sets base pre­mi­um rates by region but does­n’t nec­es­sar­i­ly take into account the actu­al soil on the farm. This could be very impor­tant and could involve a lot of work to rework work the rat­ing sys­tem, but could have some impor­tant impli­ca­tions. The fact that the pro­gram has been very suc­cess­ful in max­i­miz­ing par­tic­i­pa­tion, it’s fair to say that far less focus has been on align­ing sus­tain­abil­i­ty goals in the program.

So this rais­es the ques­tion, can we bring big data tools to bear to help align these goals in some way? So, recent­ly we did a study see­ing if we could inte­grate in high-resolution soil data to improve the rat­ing. And so by inte­grat­ing these big data sets togeth­er, we were able to come up with pre­cise field-level esti­mates of risk. It turns out that the rat­ing accu­ra­cy can be dra­mat­i­cal­ly improved by incor­po­rat­ing these data. And so omit­ting them can result in pre­mi­um errors on the order of bil­lions of dol­lars a year.

This is real­ly impor­tant because insur­ance prices can affect incen­tives in much the same way that prices can affect incen­tives in any mar­ket. So get­ting the rates right is poten­tial­ly very impor­tant. In doing a lot of work on this, we found that we could prob­a­bly rebuild the pro­gram, even though it might be some­what dif­fi­cult. But what it might enable is basi­cal­ly facil­i­tat­ing pro­mo­tion of soil health and oth­er things through crop insur­ance incen­tives, reward­ing farm­ers for improv­ing soil qual­i­ty while simul­ta­ne­ous­ly build­ing a stronger program.

If we could do this, though, why would this even mat­ter? Some peo­ple have said that hav­ing these risk man­age­ment pro­grams basi­cal­ly cre­ates dis­in­cen­tives for farm­ers to adapt to say, cli­mate change. The real­i­ty, though, is that hav­ing well-functioning cred­it risk and man­age­ment mar­kets is real­ly actu­al­ly crit­i­cal to enabling invest­ments in tech­nolo­gies to enable farm­ers to adapt.

So indeed, in the US what we found in the high­est pro­duc­tiv­i­ty regions is that we’ve seen increas­es in pro­duc­tiv­i­ty and decreas­es in risk, actu­al­ly, due to the mas­sive gains we’ve seen in biotech­nol­o­gy and man­age­ment and so forth. A lot of that is exact­ly because we have these well-functioning mar­kets to facil­i­tate that. So it’s real­ly important.

But yet despite the fact that the under­ly­ing struc­ture of the crop insur­ance pro­gram, which is the main way we sub­si­dize agri­cul­ture, is still fair­ly low-tech. Think about crop insur­ance as a con­duit to pro­mot­ing sus­tain­abil­i­ty. Some gaps must be bridged, and those gaps are inte­gra­tion of big data tools with smart pol­i­cy designs. 

The oppor­tu­ni­ties here are also not just lim­it­ed to soil health. There’s a vari­ety of areas right now where advanced com­pu­ta­tion­al meth­ods are being brought to bear to improve the func­tion­ing of these types of pro­grams. Everything from inte­grat­ing genet­ic data, to remotely-sensed data, and others. 

So for exam­ple, recent­ly we just did a study where we were able to inte­grate high-resolution genet­ic data with high-resolution large-scale yield data for all of the rice in the US grown in the last four decades. And this enabled us to do genetic-specific insur­ance pric­ing at the actu­al snip lev­el, not just on traits. Amazing advance­ments that ulti­mate­ly could help breed­ers and also improve the program.

This is also being brought to bear in devel­op­ing coun­tries, where we’re help­ing improve and cre­ate risk man­age­ment mar­kets, or we use remotely-sensed satel­lite data to basi­cal­ly struc­ture insur­ance con­tracts for pas­toral­ist live­stock pro­duc­ers. We have a lot of tools these days to do that. I mean, we can rate and price insur­ance from the genet­ic lev­el, at the soil lev­el, and even lit­er­al­ly from space, which is real­ly excit­ing to have these tools to bring to bear. But more can be done, I think, in terms of align­ing the design of these pro­grams to pro­mote some of these goals that we want to reach.

Precision agri­cul­ture and gov­ern­ment pol­i­cy and risk man­age­ment mar­kets, these all have real­ly impor­tant roles to play in both inten­si­fi­ca­tion and con­ser­va­tion in agri­cul­ture. And pol­i­cy in par­tic­u­lar has a very pro­found role in medi­at­ing this intensification/conservation dynam­ic. And advanced ana­lyt­ics sort of in some ways bring to bear the keys to unlock­ing these. 

So my ques­tion to you is what are the oppor­tu­ni­ties to bring data ana­lyt­ics to bear to improve risk man­age­ment and pol­i­cy in your mar­kets? Thank you.