cpi

A Python library that quickly adjusts U.S. dollars for inflation using the Consumer Price Index (CPI).

Installation

The library can be installed from the Python Package Index with any of the standard Python installation tools.

Like pipenv:

$ pipenv install cpi

Or pip:

$ pip install cpi

Working with Python

Adjusting for inflation is as simple as providing a dollar value followed by the year it is from to the inflate method. By default it is adjusted to its value in the most recent year available using "CPI-U" index recommended as a default by the Bureau of Labor Statistics.

>>> import cpi
>>> cpi.inflate(100, 1950)
1017.0954356846472

If you'd like to adjust to a different year, submit it as an integer to the optional to keyword argument.

>>> cpi.inflate(100, 1950, to=1960)
122.82157676348547

You can also adjust month to month. You should submit the months as datetime.date objects.

>>> from datetime import date
>>> cpi.inflate(100, date(1950, 1, 1), to=date(2018, 1, 1))
1072.2936170212768

You can adjust values using any of the other series published by the BLS as part of its "All Urban Consumers (CU)" survey. They offer more precise measures for different regions and items.

Submit one of the 60 areas tracked by the agency to inflate dollars in that region. You can find a complete list in the documentation.

>>> cpi.inflate(100, 1950, area="Los Angeles-Long Beach-Anaheim, CA")
1081.054852320675

You can do the same to inflate the price of 400 specific items lumped into the basket of goods that make up the overall index. You can find a complete list in the documentation.

>>> cpi.inflate(100, 1980, items="Housing")
309.77681874229353

And you can do both together.

>>> cpi.inflate(100, 1980, items="Housing", area="Los Angeles-Long Beach-Anaheim, CA")
344.5364396654719

Each of the 7,800 variations on the CU survey has a unique identifier. If you know which one you want, you can submit it directly.

>>> cpi.inflate(100, 2000, series_id="CUUSS12ASETB01")
165.15176374077112

If you'd like to retrieve the CPI value itself for any year, use the get method.

>>> cpi.get(1950)
24.1

You can also do that by month.

>>> cpi.get(date(1950, 1, 1))
23.5

The same keyword arguments are available.

>>> cpi.get(1980, items="Housing", area="Los Angeles-Long Beach-Anaheim, CA")
83.7

If you'd like to retrieve a particular CPI series for inspection, use the series attribute's get method. No configuration returns the default series.

>>> cpi.series.get()
<Series: CUUR0000SA0: All items in U.S. city average, all urban consumers, not seasonally adjusted>

Alter the configuration options to retrieve variations based on item, area and other metadata.

>>> cpi.series.get(items="Housing", area="Los Angeles-Long Beach-Anaheim, CA")
<Series: CUURS49ASAH: Housing in Los Angeles-Long Beach-Anaheim, CA, all urban consumers, not seasonally adjusted>

If you know a series's identifier code, you can submit that directly to get_by_id.

>>> cpi.series.get_by_id('CUURS49ASAH')
<Series: CUURS49ASAH: Housing in Los Angeles-Long Beach-Anaheim, CA, all urban consumers, not seasonally adjusted>

Once retrieved, the complete set of index values for a series is accessible via the indexes property.

>>> series = cpi.series.get(items="Housing", area="Los Angeles-Long Beach-Anaheim, CA")
>>> series.indexes
[<Index: 1997-01-01 (January): 155.4>, <Index: 1997-02-01 (February): 155.6>, <Index: 1997-03-01 (March): 155.5>, <Index: 1997-04-01 (April): 155.2>, <Index: 1997-05-01 (May): 156.1>, <Index: 1997-06-01 (June): 156.4>, <Index: 1997-07-01 (July): 156.9>, <Index: 1997-08-01 (August): 156.7>, <Index: 1997-09-01 (September): 157.1>, <Index: 1997-10-01 (October): 157.9>, ...

That's it!

Working with the command line

The Python package also installs a command-line interface for inflate that is available on the terminal.

It works the same as the Python library. First give it a value. Then a source year. By default it is adjusted to its value in the most recent year available.

$ inflate 100 1950
1017.09543568

If you'd like to adjust to a different year, submit it as an integer to the --to option.

$ inflate 100 1950 --to=1960
122.821576763

You can also adjust month to month. You should submit the months as parseable date strings.

$ inflate 100 1950-01-01 --to=2018-01-01
1054.75319149

Here are all its options.

$ inflate --help
Usage: inflate [OPTIONS] VALUE YEAR_OR_MONTH

  Returns a dollar value adjusted for inflation.

Options:
  --to TEXT      The year or month to adjust the value to.
  --series_id TEXT  The CPI data series used for the conversion. The default is the CPI-U.
  --help         Show this message and exit.

Working with pandas

An inflation-adjusted column can quickly be added to a pandas DataFrame using the apply method. Here is an example using data tracking the median household income in the United States from The Federal Reserve Bank of St. Louis.

>>> import cpi
>>> import pandas as pd
>>> df = pd.read("test.csv")
>>> df.head()
   YEAR  MEDIAN_HOUSEHOLD_INCOME
0  1984                    22415
1  1985                    23618
2  1986                    24897
3  1987                    26061
4  1988                    27225
>>> df['ADJUSTED'] = df.apply(lambda x: cpi.inflate(x.MEDIAN_HOUSEHOLD_INCOME, x.YEAR), axis=1)
>>> df.head()
   YEAR  MEDIAN_HOUSEHOLD_INCOME      ADJUSTED
0  1984                    22415  52881.278152
1  1985                    23618  53803.384387
2  1986                    24897  55682.049635
3  1987                    26061  56233.030986
4  1988                    27225  56410.752325

The lists of CPI series and each's index values can be converted to a DataFrame using the to_dataframe method.

Here's how to get the series list:

>>> series_df = cpi.series.to_dataframe()
>>>> series_df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7795 entries, 0 to 7794
Data columns (total 13 columns):
area_code              7795 non-null object
area_id                7795 non-null object
area_name              7795 non-null object
id                     7795 non-null object
items_code             7795 non-null object
items_id               7795 non-null object
items_name             7795 non-null object
periodicity_code       7795 non-null object
periodicity_id         7795 non-null object
periodicity_name       7795 non-null object
seasonally_adjusted    7795 non-null bool
survey                 7795 non-null object
title                  7795 non-null object
dtypes: bool(1), object(12)
memory usage: 738.5+ KB

Here's how to get a series's index values:

>>> series_obj = cpi.series.get(
>>>    items="Housing",
>>>    area="Los Angeles-Long Beach-Anaheim, CA"
>>> )
>>> index_df = series_obj.to_dataframe()
>>> index_df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 553 entries, 0 to 552
Data columns (total 22 columns):
date                          553 non-null object
period_abbreviation           553 non-null object
period_code                   553 non-null object
period_id                     553 non-null object
period_month                  553 non-null int64
period_name                   553 non-null object
period_type                   553 non-null object
series_area_code              553 non-null object
series_area_id                553 non-null object
series_area_name              553 non-null object
series_id                     553 non-null object
series_items_code             553 non-null object
series_items_id               553 non-null object
series_items_name             553 non-null object
series_periodicity_code       553 non-null object
series_periodicity_id         553 non-null object
series_periodicity_name       553 non-null object
series_seasonally_adjusted    553 non-null bool
series_survey                 553 non-null object
series_title                  553 non-null object
value                         553 non-null float64
year                          553 non-null int64
dtypes: bool(1), float64(1), int64(2), object(18)
memory usage: 91.3+ KB

Source

The adjustment is made using data provided by The Bureau of Labor Statistics at the U.S. Department of Labor.

Currently the library only supports inflation adjustments using series from the "All Urban Consumers (CU)" survey. The so-called "CPI-U" survey is the default, which is an average of all prices paid by all urban consumers. It is available from 1913 to the present. It is not seasonally adjusted. The dataset is identified by the BLS as "CUUR0000SA0." It is used as the default for most basic inflation calculations. All other series measuring all urban consumers are available by taking advantage of the library's options. The alternative survey of "Urban Wage Earners and Clerical Workers" is not yet available.

Updating the CPI

Since the BLS routinely releases new CPI new values, this library must periodically download the latest data. This library does not do this automatically. You must update the BLS dataset stored alongside the code yourself by running the following method:

>>> cpi.update()