It is not good idea to budget as if you are going to win the lottery

Government data shows that within the next decade the federal government may be spending more on entitlement programs and interest on the national debt than it receives in revenue. It couldn't spend money on anything else and balance the budget. Within the next 10-25 years simply the interest on the debt could be more than revenue so the country couldn't even in theory balance the budget, and would have effectively gone bankrupt   before that.

Politicians didn't learn enough from the financial crisis. Many people  lost houses when they couldn't afford rising mortgage costs. Often they  overestimated their future earnings. Its understandable to dream about what could be done with a higher income, but its best to use conservative estimates when planning budgets.  Federal agencies aren't doing a good   job of  considering the possibility of lower income and higher expenses.

People in business often plan by exploring best, expected and worst case scenarios because they know it is difficult to accurately predict the future.   The Social Security Administration (SSA) at least attempts to do that, but government-wide forecasts haven't taken that approach. Unfortunately the SSA's  estimates have enough  flaws that its unclear  whether we have a realistic estimate for how much future entitlements will cost. The Congressional Budget Office released a Long Term Budget Outlook  in June 2012 where they produced this graph of potential future debt vs. GDP (Gross Domestic Product, a $ measure of how much our economy produces):
Neither one is meant to be a worst case scenario, e.g. they both use the same forecast for future GDP even though they acknowledge they don't  take into account the possibility that high government debt levels may slow the economy. The difference is  the unrealistic baseline scenario   assumes  laws aren't changed (they are required to consider that possibility). The "alternative" scenario assumes laws are updated to continue current policies. In the baseline scenario  any expiring policies they always renew are assumed to disappear, which is how it manages to show debt eventually decreasing. Its  like doing a personal budget pretending  when your car lease expires you won't  have one and  won't have any transportation expenses, even though its not realistic since no bus runs close to your workplace 50 miles away. The "Alternative Fiscal Scenario" is based on current policies continuing. It assumes they will change laws to keep  promises they have made and renew  programs they've always kept.

For the past few decades federal revenue as a % of GDP  has only fluctuated within a small range regardless of tax rates. The unrealistic baseline scenario assumes it will somehow magically manage  to bring in more revenue than government has collected in the past. The federal GAO (Government Accounting Office) responded in December 2012 with its own version of those scenarios where it at least caps the baseline version at "21.4 percent of GDP", while the more realistic "alternative simulation" assumes revenue peaks at its "40-year historical average of 17.9 percent of GDP". Now both scenarios show debt  increasing:

Both scenarios use the same GDP forecast, which is close to the one CBO uses and the "intermediate" SSA forecast. The annual 2012 Social Security Board of Trustees Report has a lower GDP estimate it uses in its "high cost" scenario (they  also acknowledge they don't take into account the potential for high debt to reduce GDP growth so even that may be optimistic). PoliticsDebunked updated the  GAO data  to base its revenue  on  the lower SSA GDP forecast, added in the highest SSA interest rate estimate, added in estimates of the higher cost medicare scenario the SSA provides (oddly as you'll read later there weren't higher Social Security costs available to add) and caps the revenue at the realistic 17.9% rate for both scenarios. Here are the adjusted GAO scenarios vs. the originals:

In reality of course   the   eventual debt levels aren't possible (even for the original GAO current policy estimate) because well before then interest costs alone would exceed revenue. This is interest cost as a % of projected revenue:

When it hits 100% then all revenue goes to merely paying interest. That is also optimistic because the interest rate assumptions  don't vary with higher debt levels. In reality well before interest payments were at  100% of revenue the interest rate would increase because the country would be considered a bad risk, and it would more quickly reach the point where it couldn't pay  interest.   In addition any scenario where  GDP continues to grow after the government has gone bankrupt is over-optimistic. A shorter term concern is the point when spending just on entitlements (social security, medicare, medicaid, etc) and interest combined will exceed revenue, leaving no room to balance the budget with any other spending unless changes are made:


This is again being optimistic because  interest rates would climb if the US were close to that point. Even if debt weren't an issue now, forecasts show that within a couple o decades entitlements by themselves could be more than revenue which means the government wouldn't be able to cover their costs (let alone pay for anything else) and couldn't afford any interest payments to borrow to do so:

Inflation would likely makes things worse

Some people  are concerned the government might try to reduce its debt via increasing inflation. That would be a concern if the budget were balanced, but even the original GAO scenarios project a deficit every year. If you  owed $1000 and merely paid off the interest on it each year, then inflation would reduce the real (inflation adjusted) value of the principal  each year. However if you had to borrow more money each year   to cover the interest payment  then  the real value of your debt wouldn't go down. It would compound at the  real (inflation adjusted) interest rate. If inflation rates are expected to rise,  interest rates should be expected to rise even more to compensate. Higher interest rates would require the government to borrow money to cover the extra interest payment cost, which would  more than offset any reduction in the debt from inflation. Larger interest payments would lead to the point where interest payments+entitlements are more than revenue happening earlier, so overall it would make things worse.

If the government had borrowed all its debt at a low fixed rate for a long time period then inflation, e.g. like 30 years at 3%, then an inflation rate above that would provide a temporary benefit for its existing debt. Once investors discovered the government were intentionally trying to inflate away its debt they would quickly punish it by demanding an even higher rate for any future borrowing out of fear inflation would rise even faster and out of fear the US was now a poor credit risk if it was engaging in such tactics. It also wouldn't receive that much of a benefit since most of the money it has borrowed is fairly short term debt it pays off by borrowing more.The US Treasury's Office of Debt Management report for Q4 2012 shows that 26.3% of federal debt has a term of less than  a year, 39% less than  2 years, 49.1% less than 3 years, and 66.2% less than 5 years.  The government would have to  hyperinflate at an increasing rate to get much benefit, and that level of inflation would likely reduce GDP growth which would lead to even more borrowing.

For planning purposes it makes sense to not expect a hyperinflationary spiral, but to at least consider using the higher inflation rate the SSA uses in its "high cost" scenario. For rough estimating purposes its assumed inflation only impacts interest rates (even though it might e.g. negatively impact GDP). 


All of these figures are assuming the economy continues to grow. While that may be realistic, its like personal budgeting betting on getting a raise. You may be in trouble if it doesn't happen so its safest to budget as if you won't get one and then put the surplus to good use if you do. If GDP were to stay flat the next several years the point where interest by itself is more than revenue is reached even earlier:

In that case interest+entitlements are more than revenue in 2016. Studies show that as government debt grows, economic growth slows, although there is controversy over how much. Even those who are skeptical of the effect since they wish to rationalize more government spending should consider that eventually interest spending crowds out other government spending they think is important, and obviously at some point the country goes bankrupt.

A Stanford economics professor attempted to quantify the potential impact of future debt on US GDP growth using the results of the CBO long term budget outlook, the  IMF paper and this one by Reinhart,Reinhart and Rogoff.. His analysis is incomplete since it ignores  feedback effects on the IMF scenario. As GDP growth slows the government will have less revenue and so debt will grow  faster, which slows GDP growth even further. His projections use a fixed forecast of US debt. The graph below uses the original GAO estimate and current policy assumptions, it would be even worse using the alternative scenarios explored on this page. It shows    future projections for GDP   vs. the potential reduction using the methodology from the IMF paper and the Reinhart paper (the CBO document doesn't provide enough information about its methodology to adapt it):
That is  overoptimmistic because interest payments would have exceeded revenue in 2038 under the Reinart scenario and in 2039 under the IMF scenario so the GDP would likely fall even earlier. It also doesn't take into account interest rates rising due to high debt levels.  Assumption details the casual reader can skip [draft note: move this to an appendix]: the Reinhart paper projects that after total debt (including intragovernment debt) reaches 90% of GDP for 5 years (which it will have done by 2015) annual growth is reduced  by 1.2%. The IMF paper refers to a slowdown in per capita GDP growth rates of 0.16% per 10% increase in publicly held debt to GDP ratio   for medium size  debt  (60-90% of GDP) and 0.19% for higher debt levels. At the low rate of population growth forecast the total GDP growth slowdown would about the same as the figures given for per capita growth slowdown figures (it'd be a trivial amount higher since it translates into "GDPGrowthSlowdown=PerCapitaSlowdown*(1+PopulationGrowthRate) - 1"). The paper indicates that advanced economies may suffer less slowdown, but it doesn't breakout what the slowdown would be for the different levels of debt  so   the  average figures are used. That seemed an appropriate  estimating tradeoff since it also notes the impact may be nonlinear (i.e. higher for higher debt levels), and because the increased debt due to higher interest rates isn't included, so it likely winds up being overoptimistic.]

The Social Security Administration has difficulties with inflation

These scenarios may still be optimistic because they haven't taken into account the potential for higher Social Security costs. The Social Security annual report gives cost figures for every year through 2090 in nominal $  (i.e. non-inflation adjusted $) and as a % of GDP. In both cases the high cost scenario appears on the surface to actually be the high cost scenario. e.g. in nominal $ for the OASDI (Old Age Survivors and Disability Insurance) Trust fund, i.e. Social Security Insurance:


What matters however are the real (inflation adjusted) costs. You'll notice  they  only give one table  that uses real $ rather than nominal or % GDP figures. They only give high cost real $ figures through 2026, but by then the figures already show  the "high cost" scenario has become the low cost scenario. Using their %GDP figures and real GDP growth figures, when adjusted for inflation the high cost scenario turns out to be really the low cost scenario:



That is effectively adjusted using their estimates of the GDP deflator for inflation. If the CPI they use in their real cost table were used,   the difference would be even greater. Since the "high cost" scenario has lower real costs,there was no point in adding it to the adjusted GAO government wide estimates. The "low cost" scenario" is lower for the first few years, and has a higher GDP assumption so it wasn't relevant to try to add that either. The low cost scenario has a higher population and a higher labor force participation rate. Yet  even the cost per capita is higher for the "low cost" scenario:
Its true that the "high cost" scenario has the lowest income so you might wonder if the scenario is designed to be the "highest shortfall cost". That isn't the case. Inflation is actually good from a real cost perspective. They explain that

(1) the effect on taxable payroll due to a greater rate of increase in average wages occurs immediately; and (2) the effect on benefits due to a larger COLA occurs with a lag of about 1 year. As a result of these effects, the higher taxable payrolls have a stronger effect than the higher benefits, which results in lower cost rates.

So high inflation leads to lower real costs, and leads to the "trust fund" lasting longer. Despite that  they assign the high cost scenario the high inflation rate which costs less. Everything indicates they are using a rather useless  conception of "high cost" as being high nominal cost, rather than "highest possible real cost" or "highest possible real cost on top of the trust fund" or "highest real cost per capita".

They also exhibit other basic problems dealing with inflation which call into question whether anyone with much financial knowledge was ever enlisted to check these forecasts.  In a document where they describe their economic assumptions they say : "The real interest rate (real effective annual yield) on the special public debt obligations issuable to the trust funds for a given year is defined as the nominal effective annual yield less the increase in the CPI-W for the first year after issue.". The problem is that isn't accurate.  Many non-professionals don't realize the equation"RealRate=NominalRate-InflationRate" is only an approximation, some "math-light" sources never mention the real equation.

Even wikipedia explains (or a Federal Reserve paper if you prefer), that when given an inflation rate and a nominal interest rate, simple algebra shows the "real interest rate" is calculated using "RealRate=((1+InflationRate)/(1+NominalRate)) - 1". Over a long range forecast of 78 years  that can add up to e.g. a 7-8% difference in resulting interest using some of the  figures in the SSA forecast. It is trivial for a professional doing a forecast to take a few more seconds to put the exact formula into a spreadsheet or program code.  Instead e.g. in the long run for the high cost scenario they show a nominal interest rate of 6.2% minus an inflation rate of 3.8% leading to a real interest rate of 2.4%, rather than the  actual 2.31% figure.

 Its true  the uncertainty in   estimates the math is applied to (and the choice of inflation measure) will  vastly overshadow that sort of difference. The concern  is that the repetition for many years of simple errors easy to find like that  raises a  red flag regarding the potential quality of the rest of the forecast. There is no excuse for not taking a few seconds more to apply the right  math to estimates to avoid making forecasts even less accurate. Its like saying "2+2=5 is close enough since we guessed at the 2 figures".  The CBI forecast makes the same mistake, the GAO forecast doesn't provide enough information to determine if they do.  SSA makes the same conceptual error when they simplistically introduce this concept: "The annual real wage differential is defined as the annual percentage change in the average OASDI covered wage minus the annual percentage change in the CPI".

  Estimates for some figures in the  Social Security forecast appropriately use the Consumer Price Index since laws link some of their expenditures to it. They use   the GDP deflator appropriately when referring to GDP growth. It isn't always straightforward how to combine two differing measures of inflation  in the same forecast. Is unclear without more information if they   mixed the two appropriately when creating their forecasts.  e.g. the World Bank provides data on international real interest rates where: "Real interest rate is the lending interest rate adjusted for inflation as measured by the GDP deflator.", yet the SSA uses the CPI to calculate "real interest rates".  It  isn't straightforward to determine how to mix inflation rates (and there are other possible choices like the PCE to consider) and there aren't enough details provided to determine if the Social Security forecasts do so appropriately.

Social Security forecasts have   unrealistic estimates ignoring trends

One  important factor regarding social security is forecasting how many people will be in the workforce to support retirees. The "age adjusted" labor force participation rate for men has been going down steadily for the years data has been collected:


Yet if you compare the last part of that trend with their forecast for the "long term" rates they expect them to reach several decades out:


They've had to continually lower their estimates over time. We are now in the "long range" future of their 1980 forecast. For a while they appear to have noticed the downward trend, but then seem to have become overoptimistic again. This shows their long term predictions vs. the actual participation rate for the year the forecast was made:


They provide rationalizations behind the claims of a higher participation rate in their document but they seem to ignore factors which have been leading  to a lower participation rate. Other aspects of their forecast may exacerbate the trend,  e.g. if their overoptimistic estimates of future salaries are a reality then people may need to work  fewer years. The documentation of their methods   seems to show them "cherry picking" the data to from 1994-2008 to use a flatter trend. They questionably look at different age groups separately without considering that their trends may interact.  If you  use all years of available data to look at trends for  each age group separately ("trend on all data" below)  rather than looking at the overall trend it may be possible to rationalize the participation rate not going quite as low as the apparent trend. but their forecast seems a bit,er,  questionable:


An interactive graph is included below where you can explore different scenarios by changing the assumptions.  A description of the controls is included below it, you can choose the basic values to show without needing to explore the detailed options.  The graph is setup initially to show interest+entitlement spending as a % of revenue. The initial scenario 1 settings correspond to GAO's original forecast, and Scenario 2 settings correspond to the "adjusted values" using the worst cast SSA GDP, interest, and inflation, etc. Currently government estimates are given as static graphs in charts. They could instead be given as interactive "what if" tools like the graph below which would allow lawmakers, policy analysts and concerned members of the public to  explore different options themselves. Whenever lawmakers or regulators propose making changes they should provide such tools to allow their proposals to be compared to current estimates. 

The first section of controls lets you choose what to show, e.g. whether to show the baseline and/or current policy figures. You can look at 2 different scenarios,or compare 2 different values for 1 scenario (e.g. look at GDP vs. Debt). If you plot two different values you can choose to show them on the same axis, or more   use 2 different axes (e.g.
 to see interest rate vs. debt).

You can select which elements of spending are added in,  grouped into broad categories based on the GAO estimates of entitlements (Medicare, Social Security, Medicaid, etc), interest, and other spending.  

The GDP menu lets you select either the base GAO estimates for GDP, the lowest SSA estimate, GDP staying flat at  its initial level, or GDP per capita staying flat  (so the total GDP rises as population rises). The Social Security report medium cost scenario has population estimates that roughly match the GAO estimates. The SSA high cost scenario has a lower population growth estimate which is used when the "lower population" box is checked.

The "Medicare High Costs" checkbox indicates whether to add in the "high cost scenarios" from the trustees reports for Medicare (adding it to both current and baseline figures for that).

The "Alt." checkboxes allow you to see where the  spending differences arise between the  "baseline" and "current policy" figures in the GAO estimates. The current policy figures have higher spending in 2 different areas compared to the "baseline" figures: medical costs (for Medicare, Medicaid, etc) and then other costs. Checking those boxes adds them in to the "baseline" figures.  The GAO baseline scenario over the long run has a revenue of 17.9% while the current policy scenario rises to 21.4%. Checking the "cap revenue" box will cap reduce the maximum revenue for both figures to whatever is specified (though it won't raise the rates, e.g. specifying 19% will lower the baseline  revenue to 19%, but leaves the current policy revenue at 17.9%).

The "Debt Impact" button specifies whether the given GDP is lowered using the strategies from the IMF or Reinhart papers and what values to use for the GDP growth reductions. It is assumed the GDP forecasts were made for the current debt level, an reduces the GDP for debt above that level.

You can choose which  real interest rate forecast to start with. They grow for a few years and then remain at peak values for the rest of the forecast. The GAO forecast peaks at 3.14% (their nominal high rate is 5.2%, deflated by their 2% GDP deflator rate), the SSA Medium forecast at 3.22% and highest at 3.65% (their nominal rates deflated by their GDP deflator).   The Treasury department provides monthly figures for the average interest rate on the publicly marketed national debt from 2000 on. The real interest rate (based on quarterly BEA GDP deflator figures) peaked in April 2001 at 4.886%, dropping to an average of 0.38% by 2004. The "reverse history interest" scenario in the graph below assumes the interest rate could potentially reverse course just as quickly.  The "Reverse Historical" forecast uses the GAO rate for 2012 and then uses those figures in reverse, peaking at 4.886% from then on. The SSA provides historical monthly figures for their interest rates. Deflating them to real interest rates, they peaked at 9.09% in 1984 and dropped over the next 20 years to 0.92% in 2004. The "Reverse SSA History" interest rate uses those rates in reverse, then remains at the peak value for the rest of the forecast.

The inflation rate options for the GDP deflator peak at 2% for GAO, 2.2% for CBO, 2.4% for SSA Medium, and 3.3% for SSA Highest.

There is controversy over how much increased government borrowing will raise interest rates.  A simplifying assumption (likely unwarranted) was made that the basic interest  forecasts would be accurate for the current debt and deficit levels and the rates only rise when borrowing rises above those levels. Most work estimates the change based on either basis points per rise in the debt-GDP ratio or basis points per rise in the deficit-GDP ratio.  A paper by an economics prof who is Dean of the Columbia graduate school of business provides a theoretically derived value of 23.7 basis points per 10% increase in the  Debt-GDP ratio which appears to be conservative based on the data provided, while another paper suggests  30-50 basis points, and elsewhere  30-40. Estimates for the rise in interest rate per 1% increase in deficit/GDP ratio vary for instance from 25 to 19-45 BP. Another paper suggests that after an increase of 1% in the deficit to GDP ratio the interest rate constantly rises each year in  fashion (which makes sense given debt is accumulating each year), the "cumulative" scheme uses their figures (continuing the trend their data implies beyond the 10 year example they give).

 If US finances get worse its likely the rate would eventually increase   faster than these estimates since they project rates are still compartively low still at the point where interest spending alone consumes all revenue. The estimates for how much interest rates rise can be applied in one of two ways. The "Current year" option assumes that the average interest rate on the debt is increased based on the debt level at the beginning of the year, which assumes implicitly that when the money was borrowed in the past the future debt's effect on the rate was built in.. In reality some of that debt was borrowed years before that so the interest rate might not actually reflect that level of debt. The "year borrowed" assumption assumes the increase applies to the year the money is borrowed. The most recent Treasury "maturity profile" plan is used to determine when to add the extra interest costs (the profile for 2022 isn't much different from 2012's and is assumed to continue).

For some time scales of course the graph represents something that couldn't happen in the real world, e.g. interest spending that costs more than revenue since the country would be bankrupt by then. When the GDP falls too much, some figures skyrocket so you may need to shorten the timescale usefully see certain results. The GDP is given a floor of $1, which of course also couldn't happen in real life but helps keep the scales in some graphs viewable instead of letting the GDP go even further down into meaningless negative values.

Troubleshooting if the graph doesn't display:
The graph uses Micsosoft's Excel Web app and Skydrive, which have sometimes been unavailable or overloaded lately so if you don't see the graph appear you might try again later. If the graph is visible and the line below it says the status is "Ready for Input." you should be fine. Microsoft times out your connection after a few minutes. This page keeps trying to reload the graph but gives up after a few minutes, you can click reload to do it manually. If the graph isn't visible you may try clicking "reload".  Sometimes Microsoft only displays an error to the user and this page doesn't know it needs to reload the graph. Up to date browsers seem to work, but if you have an older browser version and the graph never shows then try a newer/different one. (e.g. it seems to require Internet Explorer 9 or later which isn't available for Windows XP, but works with Firefox under XP).