New earnings process

[codykallen]

The current approach to forecasting earnings should be improved to begin with BEA data and CBO forecasts, make adjustments for relevance by firm type (and industry, in the future), and then adjusted to align with IRS historical results. I expect the methodology to proceed as follows (with possible changes):

1. Begin with BEA data on corporate and noncorporate profits before tax, and use the forecasts for these from CBO.
2. Add back depreciation and net interest to get gross profits.
3. Use IRS SOI tables to split corporate profits into C corporations, S corporations, RICs and REITs, and to split noncorporate profits into farm sole proprietorships, nonfarm sole proprietorships and partnerships.
4. Use IRS SOI tables to split C corporation profits into those held by firms in taxable and nontaxable status.
5. Make adjustments to resemble actual gross profits for each firm type, in particular regarding real unreported income.

Note that this may involve splitting the Corporation class into those with positive net income and with negative net income. This would be further complicated by changes in the tax base, which could be problematic.

Another potential approach would be to begin with IRS information for 2013 and explicitly forecast each component of it. These would include

• Business receipts (sales, etc.)
• Interest income (taxable and tax-exempt)
• Rents and royalties
• Capital gains and losses
• Dividends from domestic corporations
• Dividends from foreign corporations (handled by the CFC and DomesticMNE classes)
• Other receipts
• Cost of goods sold
• Compensation of officers
• Salaries and wages
• Repairs
• Bad debts
• Rent paid
• State and local taxes paid
• Interest paid
• Charitable contributions
• Amortization (should be handled by Asset class, mostly)
• Depreciation (handled by Asset class)
• Depletion
• Advertising
• Pension plans, profit-sharing, etc.
• Employee benefit programs
• Domestic production deduction (already done)
• Net operating loss deductions
• Other deductions

All of these could be set to grow at the same (or separate) rates, unless it is explicitly modeled.

Of these approaches, I think the latter is likely better, as we have explicit initial values for 2013.

[codykallen]

The second approach described above has now been implemented by PR #106. This is only applied for C corporations in with net income (not net loss).

A simplified version could potentially be implemented for the PassThrough class, although the components of income are less detailed. For partnerships and S corporations, we have similar information on deductions, but less detailed information on revenues. For nonfarm sole proprietorships, we have richer information on deductions, but no detail on revenues. For farm proprietorships, the most recently available data are from 2004, so I do not think we can model this sector.

[codykallen]

This has been resolved by #106.