Working with Property Price Indices • realestatebr

Introduction

Brazil has several Residential Property Price Indices (RPPI) from different sources, each with unique methodologies and coverage. This vignette shows how to access and analyze these indices using the realestatebr package.

library(realestatebr)
library(dplyr)
library(ggplot2)

# Helper function for consistent plot styling
theme_series <- function() {
  theme_minimal() +
    theme(
      panel.grid.minor = element_blank(),
      legend.position = "bottom",
      palette.colour.discrete = c("#27708CFF", "#85A693FF", "#A6511FFF", "#BFA575FF", "#B4D9CEFF")
    )
}

Available RPPI Datasets

Individual Indices

Indexes from different sources can be accessed individually using the ‘table’ argument from the get_dataset() function.

# FipeZap Index (most comprehensive - 20 cities)
fipezap <- get_dataset("rppi", "fipezap")

# IVGR - National sales index from BCB
ivgr <- get_dataset("rppi", "ivgr")

# IGMI - Hedonic sales index from FGV
igmi <- get_dataset("rppi", "igmi")

# IVAR - Rent index from FGV
ivar <- get_dataset("rppi", "ivar")

# IQA - QuintoAndar rent prices
iqa <- get_dataset("rppi", "iqa")

# Secovi-SP - São Paulo market index
secovi_sp <- get_dataset("rppi", "secovi_sp")

Stacked Indices

To facilitate comparisons, all sale and rent indices are available in stacked format. This is especially useful for comparing multiple indices across different cities or regions.

# All sale indices in one table
sale_indices <- get_dataset("rppi", "sale")
unique(sale_indices$source)
# Shows: IGMI-R, IVG-R, FipeZap, Secovi-SP

# All rent indices
rent_indices <- get_dataset("rppi", "rent")
unique(rent_indices$source)
# Shows: IVAR, IQA, Secovi-SP, FipeZap

Understanding the Data Structure

Common Columns

Most RPPI datasets share these columns:

date: reference date (YMD format)
name_muni: city or region name
index: index number (base varies by source)
chg: monthly percent change
acum12m: 12-month accumulated change
price_m2: median price per square meter (if available)

Specific Cases

QuintoAndar (IQA)

Unlike other indices, IQA provides actual rent prices instead of an index. Creating a base index from these prices is straightforward. The code below shows how to create a base 100 index for each municipality, using the first available rent price as the base (i.e. 2019/06 = 100).

iqa <- get_dataset("rppi", "iqa")

iqa <- iqa |>
  mutate(
    index = rent_price / first(rent_price) * 100,
    .by = name_muni
  )

It’s important to note that IQA went through a significant change in methodology in mid-2022, which may affect trend analysis.

FipeZap

FipeZap is the most detailed RPPI, it includes both residential and non-residential markets, rent and sales indices, and breakdowns by number of bedrooms. The full level of granularity, however, is not available for all cities.

market: residential or commercial.
rent_sale: sale or rent market.
rooms: breakdown by number of bedrooms (1, 2, 3, 4+, total).
variable: index, chg, acum12m, price_m2 (median listing price), yield (rental yield).

Examples

Regional Comparison with IGMI

The IGMI-R index from Abecip/FGV providesprice indices for major Brazilian cities. The code below compares the IGMI sale index across São Paulo, Rio de Janeiro, Belo Horizonte, and Brasília.

For most use cases, this is the most reliable (i.e. best) RPPI available. The index is based on actual transaction prices that are adjusted for property characteristics using hedonic methods and provides city-level coverage. Its primary limitation is the relatively short time series, covering only 2014 to the present.

# Get IGMI data
igmi <- get_dataset("rppi", "igmi")

# Major cities comparison
cities_igmi <- igmi |>
  filter(name_muni %in% c("São Paulo", "Rio De Janeiro",
                          "Belo Horizonte", "Brasília"))

# Plot with facets
ggplot(cities_igmi, aes(x = date, y = index, color = name_muni)) +
  geom_line(linewidth = 0.8) +
  labs(title = "IGMI Sale Index by City",
       x = NULL,
       y = "Index") +
  theme_series()

IVG-R: the ‘official’ sales index

IVG-R is the national sales index published by the Central Bank of Brazil (BACEN). It is based on median prices from actual transactions but does not include hedonic adjustments. The index offers a long time series (2001-present) but is only available at the national level.

IVG-R is often considered the “official” RPPI in Brazil and is widely used by economists and policymakers. While based on median prices (see MVG-R), the final index applies both a 3-month moving average and the HP filter to smooth volatility, making it less volatile than other indices.

ivgr <- get_dataset("rppi", "ivgr")

# Plot national IVG-R index
ggplot(ivgr, aes(x = date, y = index)) +
  geom_line(color = "#27708CFF", linewidth = 0.8) +
  labs(title = "IVG-R National Sale Index",
       x = NULL,
       y = "Index (base varies by source)") +
  theme_series()

FipeZap: the broad sales index

FipeZap is the most comprehensive RPPI in terms of geographic coverage, with indices for 50 cities. However, it is based on median listing prices rather than actual transaction prices. The index is available from 2008 to the present and includes both sales and rental markets.

Working with this dataset requires filtering for specific cities and market segments. The example below compares the residential sales index across São Paulo, Rio de Janeiro, and Belo Horizonte.

# Get FipeZap data
fz <- get_dataset("rppi", "fipezap")

# Filter for major cities - residential sale index
cities <- fz |>
  filter(
    # residential x commercial market
    market == "residential",
    # sale x rent market
    rent_sale == "sale",
    # all bedroom counts (1, 2, 3, 4+, total)
    rooms == "total",
    variable == "index",
    name_muni %in% c("São Paulo", "Rio De Janeiro", "Belo Horizonte")
  )

# Plot
ggplot(cities, aes(x = date, y = value, color = name_muni)) +
  geom_line(linewidth = 0.8) +
  labs(
    title = "FipeZap Sale Index - Major Cities",
    x = NULL,
    y = "Index",
    color = "City"
  ) +
  theme_series()

Comparing National Sale Indices

# Get stacked sale data
sales <- get_dataset("rppi", "sale")

# Filter for national indices
national <- sales |>
  filter(name_muni == "Brazil")

# Plot comparison
ggplot(national, aes(x = date, y = index, color = source)) +
  geom_line(linewidth = 0.8) +
  facet_wrap(vars(source), scales = "free") +
  labs(title = "Brazil National RPPI - Sales",
       x = NULL,
       y = "Index (base varies by source)",
       color = "Source") +
  theme_series()

To actually make the indices comparable, we need to normalize them to a common base. The code below normalizes all indices to 100 at the start of 2018 and then compares their growth over time. The total variation from 2018 to 2024 is also annotated on the plot.

comp_indexes <- national |>
  filter(date >= as.Date("2018-01-01"), date <= as.Date("2024-12-31")) |>
  mutate(
    index_2018 = index / first(index) * 100,
    .by = source
  )

total_variation <- comp_indexes |>
  summarise(
    total_variation = last(index_2018) - first(index_2018),
    date = last(date),
    index_2018 = last(index_2018),
    .by = source
  ) |>
  mutate(label = glue::glue("{source}: {round(total_variation, 1)}%"))

ggplot(comp_indexes, aes(x = date, y = index_2018, color = source)) +
  geom_line(linewidth = 0.8) +
  geom_hline(yintercept = 100) +
  geom_label(
    data = total_variation,
    aes(x = date, y = index_2018, label = label),
    hjust = 0,
    nudge_x = 30,
    show.legend = FALSE
  ) +
  scale_x_date(limits = c(as.Date("2018-01-01"), as.Date("2025-08-01"))) +
  guides(color = guide_legend()) +
  labs(
    title = "Brazil National RPPI - Sales",
    x = NULL,
    y = "Index (base varies by source)",
    color = "Source"
  ) +
  theme_series()

Alternatively, we can compare the 12-month growth rates of each index, which are directly comparable across sources. Note, however, that overall trends may point to different conclusions depending on the index used. This reflects differences in methodology and coverage across indices.

ggplot(comp_indexes, aes(x = date, y = acum12m, color = source)) +
  geom_line(linewidth = 0.8) +
  geom_hline(yintercept = 0) +
  scale_x_date(limits = c(as.Date("2018-01-01"), as.Date("2025-08-01"))) +
  scale_y_continuous(labels = scales::label_percent(scale = 100)) +
  guides(color = guide_legend()) +
  labs(
    title = "Brazil National RPPI - Sales",
    x = NULL,
    y = NULL,
    color = "Source"
  ) +
  theme_series()

IVAR: the repeat-rent index

IVAR is the rental index published by FGV, based on an unique “repeat-rent” methodology. It covers major Brazilian cities and is available from 2019 to the present. The index is based on actual rental contract prices, making it a reliable measure of rental market trends.

ivar <- get_dataset("rppi", "ivar")

ivar_cities <- ivar |>
  filter(name_muni != "Brazil")

# Major cities comparison
ggplot(ivar_cities, aes(x = date, y = index, color = name_muni)) +
  geom_line(linewidth = 0.8) +
  labs(title = "IVAR Rent Index by City",
       x = NULL,
       y = "Index (base varies by source)") +
  theme_series()

IVAR is a “raw” index, meaning it does not apply any smoothing techniques like moving averages or the HP filter. As a result, it may appear more volatile than other indices.

Applying a filter can help visualize underlying trends. The code below applies a 6-month moving average to the national IVAR index.

# National IVAR index
ivar_national <- ivar |>
  filter(name_muni == "Brazil") |>
  mutate(ma_index = zoo::rollmean(index, 6, fill = NA, align = "right"))

ggplot(ivar_national, aes(x = date)) +
  geom_line(aes(y = index), color = "#A6511FFF", linewidth = 0.5, alpha = 0.6) +
  geom_line(aes(y = ma_index), color = "#27708CFF", linewidth = 0.8) +
  labs(title = "IVAR National Rent Index (raw vs 6-month MA)",
       x = NULL,
       y = "Index (base varies by source)") +
  theme_series()

FipeZap: the broad rental index

Once again, FipeZap offers the most comprehensive geographic coverage. However, it is based on median listing rent prices rather than actual contract prices.

# Get FipeZap data
fz <- get_dataset("rppi", "fipezap")

# Filter for major cities - residential rent index

start_date <- as.Date("2021-01-01")

#
cities_rent <- fz |>
  filter(
    market == "residential",
    rooms == "total",
    variable == "acum12m",
    rent_sale == "rent",
    date >= start_date,
    !is.na(value)
  )
#
cities <- cities_rent |>
  filter(date == start_date) |>
  pull(name_muni)

cities_rent <- cities_rent |>
  filter(name_muni %in% cities)


# Plot all cities
ggplot(cities_rent, aes(x = date, y = value)) +
  geom_line(linewidth = 0.8, color = "#27708CFF") +
  facet_wrap(vars(name_muni)) +
  labs(
    title = "FipeZap Rent Index - Major Cities",
    x = NULL,
    y = "Index"
  ) +
  theme_series()

IQA: QuintoAndar Rent Prices

QuintoAndar is an online brokerage platform that provides rental price data for major cities in Brazil. Initially, the IQA was a simple median price index based on rental contract prices. However, in mid-2023, QuintoAndar adopted a hedonic pricing model that incorporates both listing and contract prices. Together with this methodological change, the index rebranded to Índice QuintoAndar ImovelWeb (IQAIW).

iqa <- get_dataset("rppi", "iqa")

iqa

FipeZap by City

cities_count <- fz |>
  filter(
    market == "residential",
    rooms == "total",
    name_muni != "Índice Fipezap",
    variable == "index",
    !is.na(value)
  ) |>
  mutate(year = lubridate::year(date)) |>
  count(year, name_muni, rent_sale) |>
  count(year, rent_sale)

ggplot(cities_count, aes(x = year, y = n, fill = rent_sale)) +
  geom_col() +
  geom_hline(yintercept = 0) +
  geom_text(aes(label = n), vjust = -0.5) +
  facet_wrap(vars(rent_sale)) +
  labs(
    title = "FipeZap Coverage Over Time",
    x = "Year",
    y = "Number of Cities",
    fill = "Market"
  ) +
  theme_series()

# Get FipeZap data
fz <- get_dataset("rppi", "fipezap")

# Filter for major cities - residential sale index
cities <- fz |>
  filter(
    market == "residential",
    rooms == "total",
    variable == "index",
    rent_sale == "sale",
    name_muni %in% c("São Paulo", "Rio De Janeiro", "Belo Horizonte")
  )

# Plot
ggplot(cities, aes(x = date, y = value, color = name_muni)) +
  geom_line(linewidth = 0.8) +
  labs(title = "FipeZap Sale Index - Major Cities",
       x = NULL,
       y = "Index",
       color = "City") +
  theme_series()

Rent vs Sale - São Paulo

# Filter for São Paulo - both rent and sale
cities <- fz |>
  filter(
    name_muni %in% c("São Paulo", "Rio De Janeiro", "Belo Horizonte"),
    market == "residential",
    rooms == "total",
    variable == "index",
    date >= as.Date("2019-01-01")
  ) |>
  select(date, name_muni, rent_sale, value)

base_index <- cities |>
  filter(between(date, as.Date("2019-01-01"), as.Date("2019-12-31"))) |>
  summarise(base_index = mean(value), .by = c(name_muni, rent_sale))

cities <- cities |>
  left_join(base_index, join_by(name_muni, rent_sale)) |>
  mutate(new_index = 100 * value / base_index) |>
  select(-base_index)

# Plot comparison
ggplot(cities, aes(x = date, y = new_index, color = rent_sale)) +
  geom_line() +
  geom_hline(yintercept = 100) +
  facet_wrap(vars(name_muni)) +
  labs(
    title = "Post-pandemic Rent vs Sale Divide",
    subtitle = "FipeZap residential index",
    x = NULL,
    y = "Index (100 = 2019 average)",
    color = "Market"
  ) +
  theme_series()

International Comparison

The package includes international RPPI data from the Bank for International Settlements (BIS):

# Get BIS international RPPI
bis <- get_dataset("rppi_bis")

# Filter for select countries
countries <- bis |>
  filter(
    reference_area %in% c("Brazil", "United States", "Japan", "Germany"),
    is_nominal == FALSE,
    unit == "Index, 2010 = 100",
    date >= as.Date("2000-01-01")
  )

# Plot comparison
ggplot(countries, aes(x = date, y = value, color = reference_area)) +
  geom_line(linewidth = 0.8) +
  geom_hline(yintercept = 100) +
  labs(title = "Residential Property Prices - International",
       subtitle = "Nominal indices, 2010 = 100",
       x = NULL,
       y = "Index (2010 = 100)",
       color = "Country") +
  theme_series()

Key Differences Between Indices

Coverage

FipeZap: 20 cities, most comprehensive geographic coverage
IVGR: National
IGMI: Major cities with hedonic adjustment
IVAR: Major cities
IQA: 2 cities (São Paulo, Rio de Janeiro)

Methodology

IVGR: Median prices from actual transactions.
IGMI: Hedonic pricing model (controls for property characteristics).
IVAR: Repeat-sales methodology.
FipeZap: Median prices from listings.
IQA (old): Median prices from rental contract prices.
IQA (new): Hedonic pricing model from both listing and contract prices.

Frequency

All indices are monthly, but historical coverage varies: - IVGR: 2001-present - IGMI: 2014-present - FipeZap: 2008-present - IVAR: 2002-present - IQA: 2019-present - Secovi-SP: 2008-present

Analysis Tips

1. Use Stacked Data for Easy Comparisons

Stacked data makes it easier to compare multiple indices:

# Get all sale indices in one table
all_sale <- get_dataset("rppi", table = "sale")

# Compare across sources
all_sale |>
  filter(name_muni == "São Paulo") |>
  ggplot(aes(x = date, y = acum12m, color = source)) +
  geom_line() +
  labs(title = "São Paulo Sale Indices Comparison")

2. Normalize Indices When Comparing Levels

Since each index has a different base period, normalize them for direct comparison:

sales <- get_dataset("rppi", table = "sale")

sales_normalized <- sales |>
  mutate(index_normalized = 100 * index / first(index), .by = source)

3. Compare Growth Rates, Not Levels

Growth rates are more meaningful than absolute index levels:

# 12-month growth comparison
sales |>
  ggplot(aes(x = date, y = acum12m, color = source)) +
  geom_line() +
  labs(y = "12-month change (%)")

Next Steps

See vignette("getting-started") for package basics
See ?get_dataset for all available tables
Explore city-level data with FipeZap
Compare methodologies across indices
Analyze rent vs sale dynamics