PTFarm prices go up. What about FIAGROs?
FIAGROs are Brazilian exchange-traded funds that channel investors' money into agribusiness. The study followed 40 of these funds, day by day, from 2021 to 2025, and compared their returns with the prices of eight farm products. The link was much weaker than expected. Explained in 12 screens.
Agribusiness is a quarter of Brazil's economy and barely on the stock exchange
To bring the two worlds together, Law 14,130/2021 created FIAGROs: funds traded on the stock exchange (B3, Brazil's exchange) that invest in agribusiness chains. Anyone who buys a unit takes part, directly or indirectly, in the sector's growth.
Data from CEPEA–Esalq/USP (a University of São Paulo research center) and from B3, cited in the dissertation's introduction. In 2021, agribusiness reached 26.6% of GDP; in 2023, 28% of employed workers.
Plenty of studies on commodities. On FIAGROs, none
So the author widened the search to agribusiness, investment and commodities. Of 3,104 documents, 56 articles from 2021 to 2025 remained. Tap each step of the funnel.
Funnel: Figure 1. Topics: Table 7 (classified by two researchers, with a third one breaking ties). Methods: Table 8. No Brazilian article appeared in the Web of Science; Brazilian studies on FIAGROs exist in broader databases, but none linked commodity prices to the performance of these funds (section 2.2).
40 funds, eight farm prices and four years of trading days
Every FIAGRO traded on B3 with available prices, from October 2021 (when trading began) to December 2025:
Next to each fund's return, the daily change in:
Funds: Table 9 (Bloomberg). Spot prices from CEPEA–Esalq; the Selic rate (Brazil's base interest rate), the Ibovespa (the main Brazilian stock index) and the dollar (Ptax rate) from Bloomberg. The model uses each day's change and up to four trading days before, in a panel with robust standard errors (Stata).
The unit price and the dividend both count
Illustrative example of the dissertation's formula (section 3.1): return = (today's price − yesterday's price + dividend) ÷ yesterday's price. The dividend counts on the day the unit starts trading without it (ex-dividend date).
Table 10. Extremes range from −86% to +169% in a single day, linked, according to the author, to market events, dividends and thin trading in some funds.
What theory predicted
The dissertation started from four hypotheses, each with a good reason behind it. Place your bets before seeing what the data said.
Of the eight farm prices, only three showed up
Rises in wheat and beef cattle were followed by slightly higher returns. Sugar went the other way. Corn, soybeans, coffee, cotton and chicken left no mark. Tap each block.
Height = largest significant coefficient of each variable in the final model (Table 12, model 4; 4,749 observations from 39 funds). Read it like this: if the price rises 1%, the fund's return changes by that many percent, on the day shown.
When it shows up, the effect comes days later and is small
Pick a price and how much it changes today. The bars show the expected change in a FIAGRO's return on each of the following trading days, according to the final model. The grey band is a fund's typical swing in a single day.
Coefficients from the final model (Table 12, model 4). The band uses the standard deviation of the funds' daily return (2.18%, Table 10). Variables without a significant effect stay at zero. Lags count trading days, not calendar days.
Prices explain less than 1% of the funds' ups and downs
Imagine a thousand parts of all the daily variation in a fund's returns over time. Farm and economic prices, together, account for fewer than ten of them. The rest is the day-to-day noise of the market.
The fund itself
About 76% of the variation in returns comes from differences between the funds themselves: what they hold, the sector they are exposed to, their protection against risk and their management strategy.
Within R² from Table 12: 0.08% (same day only, model 1), 0.83% (model 2) and 0.49% (final model). Share of the variance linked to differences between funds (ρ = 0.76): model 1, section 4.1. Dots rounded to the nearest whole number. Even though small, the set of variables is statistically significant (F test).
With a different yardstick, only sugar held
To check whether the results depended on extreme days (such as +169% or −86%), the author redid the calculations with another method, LAD (least absolute deviations): instead of adding up squared errors, it adds up errors without their sign, which gives less weight to unusual days.
Main model: Table 12, model 4 (4,749 observations). LAD: Table 14, model 3 (27,317 observations). ↑ the fund rises with it; ↓ the fund falls; "—" no significant effect. In brackets, the trading day on which the effect appears. In Table 14, the dollar's effect on the 2nd trading day has a positive coefficient and a negative t statistic; we follow the coefficient and the author's text.
They paid for the risk, with a dip in 2023 and 2024
Three ratios compare what a fund earned above the Selic rate with the risk it took. Above zero, the fund paid for the risk; below, it did not. Tap each half-year.
Tables 16 to 18. N = number of funds in the half-year. Excess return = FIAGRO return − Selic rate. Sharpe divides by the total swing; Sortino, only by the downward swing; Treynor, by beta (how much the fund follows the Ibovespa).
Investing in agribusiness through a FIAGRO is not a bet on soybean prices
For the author, FIAGROs are an asset class with dynamics of their own: only loosely tied to commodities and also to the traditional stock market.
For investors
- Being exposed to agribusiness does not mean following commodity prices.
- FIAGROs can help diversify a portfolio, behaving differently from farm markets and the stock exchange.
When looking at a fund
- What matters most is the fund itself: its portfolio, credit quality, contracts and management.
- When counting returns, dividends matter: they are the main way unit holders are paid.
For managers and regulators
- The evidence can help improve how these funds are structured.
- And help develop the capital market that finances agribusiness.
For researchers
- Split the funds by the kind of asset that dominates their portfolio.
- Use a longer time window and include credit quality, liquidity and sustainability indices.
For the author, the main finding is not the few links that showed up, but their size: much smaller than theory led one to expect from funds tied to agribusiness.
What this research does not say
- A short history. FIAGROs have existed since 2021, and the period studied is just over four years. · according to the author
- Very different funds lumped together. Credit funds, land funds and funds holding other assets, with different structures, were analyzed together. · according to the author
- Short term. The model looks at each day's change and up to four trading days before. It says nothing about months or years. · our observation
- Fewer days in the final model. With the lags, the final model used 4,749 of the 27,328 observations; the test with the other method used 27,317. · our observation, with data from Tables 12 and 14
- Little link with the stock market. This conclusion comes from the low betas that made the Treynor ratio of little use; the beta values themselves are not reported. · our observation
Credits
PALHÃO, Jessica Paganotti. FIAGROS e commodities: análise de risco e retorno entre 2021 e 2025 [FIAGROs and commodities: risk and return analysis between 2021 and 2025]. Master's dissertation (Business Administration) — Mackenzie Presbyterian University, São Paulo, 2026. Advisor: Prof. Dr. Denis Forte.
Research supported by CAPES (Brazil's federal agency for graduate education). Bibliometric and systematic review in the Web of Science (Biblioshiny), panel data regression and a LAD robustness test (Stata). According to the author, the text had language revision with the help of ChatGPT. This version simplifies the work for the general public; the full review and tables are in the original text.