Demo of CCC for PSL Meetup#

This notebook provides a demonstration of the Cost-of-Capital-Calculator (CCC) for the PSL Meetup on April 29, 2019.

To run this notebook on your machine, you will need to follow the instructions to install CCC as described in the CCC README here. In particular, you need to:

  • Install the Anaconda distribution of Python

  • Install the CCC package by typing conda install -c conda-forge ccc (or pip install cost-of-captial-calculator) in the command prompt.

Once you follow the above, you will be ready to work with this Jupyter Notebook.

First things first, import necessary packages#

# To install ccc package (if not already):
import sys
if 'ccc' not in sys.modules:
    !pip install cost-of-capital-calculator
Requirement already satisfied: cost-of-capital-calculator in /home/runner/work/Cost-of-Capital-Calculator/Cost-of-Capital-Calculator (1.3.0)
Requirement already satisfied: taxcalc in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from cost-of-capital-calculator) (3.5.3)
Requirement already satisfied: pandas in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from cost-of-capital-calculator) (2.2.2)
Requirement already satisfied: bokeh in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from cost-of-capital-calculator) (3.4.1)
Requirement already satisfied: numpy in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from cost-of-capital-calculator) (1.26.4)
Requirement already satisfied: paramtools in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from cost-of-capital-calculator) (0.0.0)
Requirement already satisfied: Jinja2>=2.9 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (3.1.3)
Requirement already satisfied: contourpy>=1.2 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (1.2.1)
Requirement already satisfied: packaging>=16.8 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (24.0)
Requirement already satisfied: pillow>=7.1.0 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (10.3.0)
Requirement already satisfied: PyYAML>=3.10 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (6.0.1)
Requirement already satisfied: tornado>=6.2 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (6.4)
Requirement already satisfied: xyzservices>=2021.09.1 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from bokeh->cost-of-capital-calculator) (2024.4.0)
Requirement already satisfied: python-dateutil>=2.8.2 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from pandas->cost-of-capital-calculator) (2.9.0)
Requirement already satisfied: pytz>=2020.1 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from pandas->cost-of-capital-calculator) (2024.1)
Requirement already satisfied: tzdata>=2022.7 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from pandas->cost-of-capital-calculator) (2024.1)
Requirement already satisfied: marshmallow>=3.0.0 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from paramtools->cost-of-capital-calculator) (3.21.1)
Requirement already satisfied: fsspec in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from paramtools->cost-of-capital-calculator) (2024.3.1)
Requirement already satisfied: sortedcontainers in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from paramtools->cost-of-capital-calculator) (2.4.0)
Requirement already satisfied: setuptools in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from taxcalc->cost-of-capital-calculator) (69.5.1)
Requirement already satisfied: numba in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from taxcalc->cost-of-capital-calculator) (0.59.1)
Requirement already satisfied: requests in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from taxcalc->cost-of-capital-calculator) (2.31.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from Jinja2>=2.9->bokeh->cost-of-capital-calculator) (2.1.5)
Requirement already satisfied: six>=1.5 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from python-dateutil>=2.8.2->pandas->cost-of-capital-calculator) (1.16.0)
Requirement already satisfied: llvmlite<0.43,>=0.42.0dev0 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from numba->taxcalc->cost-of-capital-calculator) (0.42.0)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from requests->taxcalc->cost-of-capital-calculator) (3.3.2)
Requirement already satisfied: idna<4,>=2.5 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from requests->taxcalc->cost-of-capital-calculator) (3.7)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from requests->taxcalc->cost-of-capital-calculator) (2.2.1)
Requirement already satisfied: certifi>=2017.4.17 in /usr/share/miniconda3/envs/ccc-dev/lib/python3.9/site-packages (from requests->taxcalc->cost-of-capital-calculator) (2024.2.2)
# import packages
import pandas as pd
import numpy as np
from bokeh.plotting import figure, show
from bokeh.io import output_notebook
# import CCC classes that we'll work with
from ccc.data import Assets
from ccc.parameters import Specification, DepreciationParams
from ccc.calculator import Calculator
# to print bokeh plots inline
output_notebook()
Loading BokehJS ...

Create an instance of the Assets class#

This is the class that contains the data that underlie CCC. The basic object is a Pandas DataFrame where each row represents a combination of a specific type of asset, industry (approximately 6-digit NAICS) and tax treatment. The columns represent names and codes for the asset and industry classifications, the tax depreciation rules used for that asset, and the rate of economic depreciation (delta).

assets = Assets()
assets.df.head(n=5)
Unnamed: 0 tax_treat assets bea_asset_code bea_ind_code Industry minor_code_alt major_asset_group minor_asset_group major_industry asset_name delta
0 0 corporate 0.0 RD32 110C Farms 111 Intellectual Property Intellectual Property Agriculture, forestry, fishing, and hunting Aerospace products and parts manufacturing 0.22
1 1 corporate 0.0 RD32 113F Forestry, fishing, and related activities 113 Intellectual Property Intellectual Property Agriculture, forestry, fishing, and hunting Aerospace products and parts manufacturing 0.22
2 2 corporate 0.0 RD32 113F Forestry, fishing, and related activities 114 Intellectual Property Intellectual Property Agriculture, forestry, fishing, and hunting Aerospace products and parts manufacturing 0.22
3 3 corporate 0.0 RD32 2110 Oil and gas extraction 211110 Intellectual Property Intellectual Property Mining Aerospace products and parts manufacturing 0.22
4 4 corporate 0.0 RD32 2120 Mining, except oil and gas 212110 Intellectual Property Intellectual Property Mining Aerospace products and parts manufacturing 0.22

Create instances of the two parameters classes#

The Specification class contains many of the model parameters, although depreciation system parameters are contained in the DepreciationParams class. Both are required arguments for the Calculator object.

A Specification object has methods that load the data from a file that contains the default parameter values (default_parameters.json) and then stores them as attributes of the Specification class object.

p.u looks into the instance of this class named p and executing the cell below will show you that p.u is dictionary containtin marginal tax rates on corporate and pass-through income.

Note that the Specification class has some arguments, such as call_tc, which defaults to False, but if set to True will call the Tax-Calculator to estimate marginal tax rates on individual filers’ income.

Similar to Specification, DepreciationParams loads default parameters from a JSON file, in this case the file tax_depreciation_rules.json.

# Create an instance of the Specification class
# p = Specification()
p = Specification(call_tc=True)
# Look at attributes
p.u
Calculator initial year =  2014
Calculator year =  2024
year:  2024
{'tau_pt': array([0.20196805]), 'tau_div': array([0.17450677]), 'tau_int': array([0.31306305]), 'tau_scg': array([0.28481273]), 'tau_lcg': array([0.18401915]), 'tau_td': array([0.2059908]), 'tau_h': array([0.03463809])}
{'c': array([0.21]), 'pt': array([0.20196805])}
# Create an instance of the DepreciationParams class
dp = DepreciationParams()
# Look at an attribute of the dp object
dp.asset[0]
OrderedDict([('value', {'life': 5.0, 'method': 'DB 200%'}),
             ('asset_name', 'Mainframes'),
             ('BEA_code', 'EP1A'),
             ('minor_asset_group', 'Computers and Software'),
             ('major_asset_group', 'Equipment'),
             ('ADS_life', 5.0),
             ('GDS_life', 5.0),
             ('system', 'GDS'),
             ('year', 2020)])

Create an instance of the Calculator class#

This class does the calculations on the data. It takes as arguments the data object (named assets here) and the parameters object (named p here).

Excuting the cell below creates an instance of the calculator class with these data and parameters, but does not yet excecute any calculations.

# Create an instance of the Calculator class
calc1 = Calculator(p, dp, assets)

With an instance of the Calculator class created, we can start doing some calculations with these data.

For instance, we can compute a table of the share of corporate vs non-corporate assets across each industry.

# Look at shares of assets across industry
calc1.asset_share_table()
Industry Corporate Pass-Through
0 Agriculture, forestry, fishing, and hunting 0.309688 0.690312
1 Mining 0.826839 0.173161
2 Utilities 0.948540 0.051460
3 Construction 0.286840 0.713160
4 Manufacturing 0.859014 0.140986
5 Wholesale trade 0.715166 0.284834
6 Retail trade 0.728628 0.271372
7 Transportation and warehousing 0.789652 0.210348
8 Information 0.918722 0.081278
9 Finance and insurance 0.786056 0.213944
10 Real estate and rental and leasing 0.212052 0.787948
11 Professional, scientific, and technical services 0.501699 0.498301
12 Management of companies and enterprises 0.824397 0.175603
13 Administrative and waste management services 0.569110 0.430890
14 Educational services 0.590168 0.409832
15 Health care and social assistance 0.510425 0.489575
16 Arts, entertainment, and recreation 0.490811 0.509189
17 Accommodation and food services 0.482236 0.517764
18 Other services, except government 0.343467 0.656533

Specifying a reform policy#

To see some more interesting results, we will want to create another Calculator object with a change in policy or economic assumptions. We can do this in a way analogous to our original instantiation of the Calculator object above. In particular, we’ll need to create a new Specification object (we can use the same underlying data, which was in the object we named assets).

In the code below, we’ll specify our “reform” as current law tax policy for 2026 (the baseline parameters above came from the default model year, 2019). In addition, we’ll increase the corporate income tax rate from 21 to 25%.

# Create another policy
p2 = Specification(year=2026)
p2.update_specification({'CIT_rate': 0.35})
calc2 = Calculator(p2, dp, assets)

Tabular output#

Now with two Calculator objects named calc1 and calc2 (representing the baseline and reform policies), we are ready to compute some of the changes in effective tax rates, cost of capital, or other variables measured in this model.

We start with an overall summary table showing the marginal effective total tax rates (METTRs) for all investments, corporate investments, and pass-through investments under varying financing assumptions. This is done through the summary_table function. It takes a calculator object as an argument.

# Look at differences in METTRs between the two policies
calc1.summary_table(calc2) # calc1 is the baseline, calc2 the reform
Marginal Effective Total Tax Rate Under Baseline Policy Marginal Effective Total Tax Rate Under Reform Policy Change from Baseline (pp)
0 Overall 18.633307 34.227392 15.594084
1 Corporations 19.280286 42.283875 23.003589
2 Equity Financed 22.115071 46.012449 23.897378
3 Debt Financed 9.093596 26.806619 17.713023
4 Pass-Through Entities 19.371175 22.561638 3.190463
5 Equity Financed 17.677259 22.481049 4.803790
6 Debt Financed 25.753921 23.022849 -2.731072

the Calculator.summary_table() method defaults to showing the results for the METTR, but there is a keyword argument that would allow you to view the output for other variables computed in CCC. We can use this to see changes in the cost of capital (denoted by \(\rho\) in the model):

# Look at how the cost of capital changed
calc1.summary_table(calc2, output_variable='rho')
Cost of Capital Under Baseline Policy Cost of Capital Under Reform Policy Change from Baseline (pp)
0 Overall 5.778887 7.149009 1.370121
1 Corporations 5.825206 8.146925 2.321719
2 Equity Financed 6.721989 9.697451 2.975462
3 Debt Financed 3.925764 4.875811 0.950047
4 Pass-Through Entities 5.723676 5.959491 0.235816
5 Equity Financed 6.359623 6.753724 0.394101
6 Debt Financed 4.169671 4.021736 -0.147936

One can also save results to disk by specifying an output type (‘excel’, ‘json’, ‘csv’, ‘tex’) and a file path:

# Save these results to disk
calc1.summary_table(calc2, output_variable='rho', output_type='excel', path='cc_table.xlsx')

There are also Calculator methods to compute summary tables by asset type or industry. These are computed in the next two cells.

# Summary by asset type
calc1.asset_summary_table(calc2)
Category Marginal Effective Total Tax Rate Under Baseline Policy Marginal Effective Total Tax Rate Under Reform Policy Change from Baseline (pp)
0 Overall 18.633307 34.227392 15.594084
1 Corporate 19.280286 42.283875 23.003589
2 Equipment 12.258208 51.215461 38.957253
3 Structures 19.422595 37.093725 17.671130
4 Intellectual Property 9.368992 52.580800 43.211808
5 Inventories 28.793133 39.340832 10.547699
6 Land 26.047757 35.224690 9.176933
7 Pass-through 19.371175 22.561638 3.190463
8 Equipment 7.951552 14.201495 6.249944
9 Structures 17.898514 20.688527 2.790013
10 Intellectual Property 4.594527 6.197002 1.602475
11 Inventories 24.821069 28.437834 3.616765
12 Land 21.821915 24.792685 2.970770
# Summary by industry
calc1.industry_summary_table(calc2)
Category Marginal Effective Total Tax Rate Under Baseline Policy Marginal Effective Total Tax Rate Under Reform Policy Change from Baseline (pp)
0 Overall 18.633307 34.227392 15.594084
1 Corporate 19.280286 42.283875 23.003589
2 Agriculture, forestry, fishing, and hunting 20.524930 39.818858 19.293928
3 Mining 12.208656 40.327504 28.118849
4 Utilities 12.003855 35.609794 23.605939
5 Construction 20.678931 44.145930 23.466999
6 Manufacturing 18.972863 44.969233 25.996370
7 Wholesale trade 24.073145 44.169123 20.095978
8 Retail trade 24.731085 41.374044 16.642959
9 Transportation and warehousing 13.646318 39.116628 25.470310
10 Information 15.057767 45.620258 30.562491
11 Finance and insurance 20.420315 48.917858 28.497543
12 Real estate and rental and leasing 24.416360 37.281611 12.865251
13 Professional, scientific, and technical ser... 16.872451 52.639298 35.766847
14 Management of companies and enterprises 22.935419 45.744359 22.808940
15 Administrative and waste management services 17.120884 48.192344 31.071460
16 Educational services 22.738930 40.851837 18.112906
17 Health care and social assistance 21.563559 42.539326 20.975768
18 Arts, entertainment, and recreation 17.683288 40.545762 22.862475
19 Accommodation and food services 24.348853 40.399439 16.050585
20 Other services, except government 21.329918 39.916087 18.586169
21 Pass-through 19.371175 22.561638 3.190463
22 Agriculture, forestry, fishing, and hunting 17.574303 21.690466 4.116164
23 Mining 8.117934 11.042248 2.924314
24 Utilities 7.838669 12.515477 4.676809
25 Construction 17.506934 21.279749 3.772815
26 Manufacturing 15.356720 18.388662 3.031942
27 Wholesale trade 21.538466 25.269057 3.730591
28 Retail trade 21.620148 25.163689 3.543541
29 Transportation and warehousing 9.838366 13.453040 3.614675
30 Information 11.121542 17.325552 6.204010
31 Finance and insurance 19.166477 22.768312 3.601835
32 Real estate and rental and leasing 20.941365 23.861741 2.920376
33 Professional, scientific, and technical ser... 12.593293 16.557602 3.964309
34 Management of companies and enterprises 18.681588 22.019487 3.337899
35 Administrative and waste management services 12.809706 17.444610 4.634904
36 Educational services 18.445926 21.470365 3.024439
37 Health care and social assistance 17.285400 20.851759 3.566359
38 Arts, entertainment, and recreation 13.230066 17.263597 4.033532
39 Accommodation and food services 20.253049 23.651422 3.398374
40 Other services, except government 16.829902 19.812388 2.982487

Visualizations#

If one wants to visualize the effects of changes in tax policy, the Calculator class has a few methods for this.

We can use the grouped_bar method to show differential effects across assets (the default):

# Visualizing changes by asset type
aplot = calc1.grouped_bar(calc2)
show(aplot)

or we can change from the default asset category split to a split by industry by changing the value of the group_by_asset keyword argument:

# Visualizing changes by industry
iplot = calc1.grouped_bar(calc2, group_by_asset=False)
show(iplot)

There’s also a plot that illustrates the range of the effects of taxes on investments across asset types, by showing the min, max, and mean values:

# Plot to show variation in METTRs across assets
rplot = calc1.range_plot(calc2, output_variable='metr')
show(rplot)

changing to show the effects on pass-through businesses and for a different output variable:

show(rplot)
# Plot to show variation in METTRs across assets
output_notebook()
rplot2 = calc1.range_plot(calc2, corporate=False, output_variable='mettr')
show(rplot2)
Loading BokehJS ...

There’s a bubble plot too, though it doesn’t yet allow for many options (currently only plots METTRs for corporate entities):

bplot = calc1.asset_bubble(calc2)
show(bplot)
WARNING:bokeh.core.validation.check:W-1000 (MISSING_RENDERERS): Plot has no renderers: figure(id='p1397', ...)
WARNING:bokeh.core.validation.check:W-1000 (MISSING_RENDERERS): Plot has no renderers: figure(id='p1458', ...)

Summary#

This notebook provides a brief example of how one would work with CCC. Please explore the source code to see additional flexiblity in the functions. And please leave any questions or suggestions in the CCC repo at PSLmodels/Cost-of-Capital-Calculator#issues.