The Washington Post put out an oped on the partial privatization of Social Security over the weekend, which Kevin Drum appropriately jumped on.
One item in the oped jumped out at me:
The second risk in personal accounts is that their transition costs might scare financial markets. As we have argued before, investors shouldn’t take fright. The transition borrowing merely swaps government debt to future retirees for government debt to bondholders. Indeed, if personal accounts are coupled with a benefit cut, as they should be, the government’s total debts would fall; if anything, investors should be heartened. But financial markets are not perfectly rational. Coming on top of the president’s irresponsible budget policies, a flood of new bond issues might possibly frighten investors, perhaps triggering a fall in the dollar and higher interest rates.
While the Post oped writers include some appropriate caveats at the end, they basically endorse the White House argument that “debt issued to finance Social Security privatization should not count” because it will be offset by future benefits cuts, and thus will reduce the government’s implicit liabilities.
You have gotta be kidding me.
Let me tell you why:
1) If the markets care about implicit liabilities, they are not showing it. The prescription drug benefit, as Paul Krugman and others (including many conservatives) have pointed out, created a larger “implicit liability” than the current “implicit liability” associated with the Social Security system (the funding gap that arises from a roughly 1.5% of GDP gap between promised benefits and expected revenues after 2052, per the CBO, and after 2042, per the Trustees — the gap is less than 1.5% of GDP in 2052, but it rises over time). I don’t think there is much evidence that the markets penalized the US government for the increase it the government’s implicit liability.
2) Best I can tell, using realistic forecasts, the consolidated government has a gap between its projected revenues and its projected expenditures of 3.5% of GDP from now until eternity, and probably more (expenditures start rising sharply after 2010). Take Social Security’s cash flow surplus – including the interest income from the Trust Fund that Social Security reinvests in governments bonds – out of the consolidated budget and that gap is well over 4% of GDP for the next few years, indeed close to 5% of GDP. The general government’s cash flow deficit, assuming no change in current policies, has a much higher present discounted value than the gap in the Social Security system: the gap is bigger than the gap in Social Security, and it starts today, not in 2052. If the markets want to worry about something, they have plenty of things to worry about that “hit” long before the Social Security funding gap emerges in 2052.
3) No one seems that worried now, whether about the current cash flow deficits of the general government or the implicit liablities created by promised Social Security and Medicare benefits. American investors seem to think that the government will eventually solve its fiscal problems (including problems that are much, much more urgent than the problems in Social Security), and will come up with the revenues to pay its existing stock of bonds without inflating away their real value. I hope they are right, though I suspect that it will take a bit of pressure from the bond market to bring about the needed fiscal policy changes. I don’t quite understand what foreign buyers of longer-term Treasuries are thinking. It is pretty clear that the US dollar will have to fall substantially over the next ten years (in real terms) to bring the trade deficit down, so the real value of their bonds will fall. At least for now, many central banks seems to believe future capital losses are an acceptable price to pay for current export growth. No guarantee this will last though. One risk of Bush’s proposed reform is pretty easy to see: the US might start flooding the market with Treasuries just when the biggest current buyers of Treasuries, the world’s central banks, start to pull back …
4) The notion that the ten year bond has already “priced in” the gap between Social Security revenues and expenditures that develops after 2052 always struck me as a bit ludicrous. Holders of ten year bonds have to worry about the risk that interest rates may rise over the next few weeks, or next few months, or next few quarters. Any rise in interest rates reduces the current market value of bonds — a concern of active traders. Buy and hold investors in the US should worry about the risk that inflation may pick up and erode the real value of their bond between now and 2015; buy and hold investors abroad should worry about the risk of dollar devaluation. Since Social Security payroll taxes exceed expenditures until 2018, the last thing anyone holding a ten year bond needs to worry about is Social Security; the only risk here is that W may come up with enough support for his plan to take money out of the system and flood the market with Treasuries. Even holders of the thirty-year bonds issued in 2000 shouldn’t worry about the gaps between Social Security revenues and expenditures that don’t materialize until in 2052 — though they might want to worry a bit about the need to start issuing new bonds to redeem the Social Security Trust Fund’s existing holdings of bonds around 2028 if the government does not get its overall fiscal act together before then. The bonds held by the Trust fund are real, and servicing them will put pressure on the rest of the government …
5) Finally, let’s talk about the cash flows associated with Social Security reform along the lines the President seems to be talking about.
In broad terms, the current Social Security system takes in about 5% of GDP in tax revenues, and pays out about 4.25% of GDP in benefits (for more details, see the CBO spreadsheet). It also collects roughly 0.75% of GDP in interest on its existing holdings of Treasuries. After 2010 or so, the baby boom starts to retire, and benefits creep up. By around 2018, they exceed payroll tax revenue, and the Trustees will need to use the interest on their bonds to make payments. After 2028, or maybe later, depending on your assumptions, benefits will exceed payroll tax revenue and interest income, and the trustees will have to start selling some of their assets (the bonds) to cover payments. That is why have run a payroll tax surplus for the past twenty years or so — we raised the payroll tax in the 1980s so we could pay Social Security benefits to the Baby boom without raising payroll taxes from 2020 on. After 2042 (trustees), or 2052 (CBO), the trust fund will be gone. Social Security payroll tax benefits of around 5% of GDP won’t cover the estimated 6.5% of GDP cost of promised benefits. That gap lasts from 2052 on, and gets a bit wider over time, but not much.
What happens if you “reform” Social Security along the lines the president is talking about. You take 1% of GDP in revenues out of the system, maybe more (depending on how large the private accounts are), introducing an immediate gap between payroll tax revenues (now 4% of GDP) and Social Security benefits (4.25% of GDP). The system has to draw on the Trust Fund more quickly, or it needs to receive cash infusions from general revenues immediately. Social Security benefits still rise as a share of GDP as a result of the baby boom’s retirement. They go up to say 5.5% of GDP, maybe more, before starting to fall as a result of the phased in benefit cuts. That leads to sustained cash flow gaps in Social Security — substantial gaps, as the CBO’s analysis of the Commission’s plan two makes clear.
Those gaps last for the first 45 years or so of most proposed reforms, and lead to a substantial increase in the stock of outstanding US government debt in the markets. After 45 years, traditional Social Security benefits start to fall substantially in the proposed plan. In the Commission model two plan, they fall to something like 2% of GDP by 2100. It is unrealistic to think that payroll tax revenues will stay at 4% if benefits are less than that, with the difference applied to pay back the bonds used to fund the transition. It is more realistic to assume payroll taxes will gradually be reduced in line with reduced benefits, or the payroll tax will be used to fund Medicare.
What is the basic trade-off then? The reform basically creates cash flow deficits in Social Security (defined as the payroll tax, the money set aside from past payroll tax surpluses, and any interest on those surpluses) where there were none from now until 2050 or so. In exchange, the reforms theoretically get rid of the gap between promised benefits and expected revenues after 2050 (Social Security’s implicit liability).
The only certainty associated with the reform is that it will increase the amount of marketable debt in the near term, and that this increase will lead to a significant increase in the United States “debt to GDP ratio”, with debt here being “debt sold to the public.” Any offsetting cost savings are purely hypothetical — they are no more real that the CBO budget forecasts that assume the budget deficit will go away because the tax cuts expire. Any future benefit cut embedded in the reform could (and, if it is too draconian, almost certainly will) be reversed by future Congresses.
In other words, the proposal effectively takes a government program that is now fully funded, indeed over-funded, on a cash flow basis, and creates a cash flow deficit in that program immediately. After all, if the rest of the government currently operates with cash flow deficits, why shouldn’t Social Security too?
If you hold a 10 year bond, or even a 30 year bond with a residual maturity of 25 years, the cash flows of the proposed reform are negative until after your bond matures, and the reform would involve a substantial increase in Treasury issuance and the overall stock of Treasuries in the market while the bond you hold is outstanding.
What is for the bond market not to like?
(Sorry about the rant; I really do think the Post could have done a bit better on this one)
By the way, I recommend Jason Furman’s work on Social Security for the Center on Budget and Policy Priorities. He hardly needs the plug — David Wessel drew heavily on his work for a Wall Street Journal feature last week. Full disclosure: Furman is an active Democrat, and no fan of the President. He also knows Social Security inside out and produces solid numbers.