Understanding policy in the great recession: Some unpleasant fiscal arithmetic

Cochrane organizes the whole essay around the government-debt valuation equation, which says the real value of outstanding money plus government debt must equal the expected present value of future primary surpluses, together with a conventional money-demand equation (Section 1, p. 2; Section 2.1, pp. 3-4). Formally, “Mt + Bt / Pt = Et [integral of Lt+t/Lt (st+t) dt],” where M is money, B is government debt, L is the real stochastic discount factor, and st = Tt - Gt is the primary surplus (tax revenue net of spending). Cochrane calls this “not unusual or counterintuitive”: if at the current price level the real value of government debt exceeds expected future surpluses, people try to shed government liabilities for private assets and goods, an “aggregate demand” or “wealth effect of government debt” (p. 4). He stresses that “except for the small seigniorage term … there is no difference between money and bonds” in this equation, so open-market operations that merely swap money for debt have no effect on the price level on their own (p. 6) – a sharp departure from textbook MV = PY reasoning, in which only money, not total government liabilities, matters.

Cochrane argues the shock was a “flight to quality” – a surge in demand for all government debt, not just money, at the expense of private debt and of goods and services – which in the fiscal equation corresponds to a decline in the discount rate applied to government liabilities (Section 3, pp. 7-8). He rules out the classic Friedman-and-Schwartz mechanism (a money shortage relative to demand, unaccommodated by the Fed): “the Federal Reserve flooded the country with money (reserves). There is no evidence for a flight to money at the expense of government bonds,” and bank deposits rose rather than fell (p. 7). Instead, he documents (Figs. 2-5) government bond rates falling sharply while private (Baa/Aaa) rates and credit spreads rose, and argues government bonds became “practically the only security one could easily repo,” so institutions wanted more of both money and government debt at once, a “special” or “liquidity” premium that MV(·) = PY “does not allow us to address” but that the symmetric-treatment-of-M-and-B fiscal equation (10) can (pp. 8-10).

Deficit spending is stimulative if and only if people do not expect future taxes or spending cuts to pay off the added debt – and because U.S. debt is nominal, this can be a fully rational expectation, not a case of “Ricardian equivalence” failure requiring irrationality or market incompleteness (Section 4, pp. 12-13). “It can be perfectly rational for people to expect that the government does not plan to raise future surpluses,” Cochrane writes, contrasting this with standard Ricardian-equivalence analysis, which assumes real debt that must eventually be repaid. A further consequence is that the common objection that stimulus “will not stimulate in time” because spending is disbursed slowly does not apply in this analysis: “future deficits (st+t for large t) are just as effective as current deficits, and possibly more so” – what matters is convincing people that future deficits will not be offset by future surpluses, which Cochrane argues is in practice very hard to communicate credibly, since governments’ debt-issuing institutions are built to signal the opposite (Section 4.2-4.3, pp. 13-14).

Cochrane distinguishes three kinds of Fed asset purchases and argues they have very different effects: pure reserve-for-short-term-Treasury swaps do essentially nothing once rates are near zero; purchases of long-term debt can shift the timing of inflation but not its total magnitude; and purchases of private debt can help only by relieving a genuine liquidity premium for government debt over private debt, an effect that necessarily runs out once that premium is satisfied (Section 5.1.2-5.1.4, pp. 15-16). On the first, “trading M for B, especially at zero rates, is like trading green M&M’s for red M&Ms. The color on your plate might change, but it would not help your diet” (p. 15). On the second, buying long-term debt “transfers inflation (aggregate demand) from the future to now, and lowers long-term nominal rates,” but “can only rearrange the timing, but not the level of deflation” (p. 15) – and he argues the more stabilizing policy, given the fiscal-shock-absorbing role of long debt (Q6), would actually be to lengthen the government’s debt maturity structure, the opposite of “operation twist”-style purchases. On the third, once “any special demand for government debt is satiated, then exchanges of private debt for government debt have no effect at all in a fiscal analysis” (p. 16).

Cochrane argues there is no substantial economic impediment: the Treasury can simply issue new debt to soak up outstanding reserves even if the Fed itself holds mostly illiquid private assets, and raising the interest rate on reserves achieves the same effect “with the stroke of a pen” (Section 6.1.2-6.1.3, pp. 17-18). He directly rebuts the claim (citing Laffer, 2009) that a trillion-dollar reserve unwind would compete with Treasury’s own new issuance and cause “failed auctions” and falling bond prices: “Prospective investors in new government debt were already holding currency or reserves, which are just a different maturity of government debt. It takes almost no additional fiscal resources to unwind a reserve or currency expansion” (p. 18), since seigniorage on even $1 trillion is only about $10 billion. He does allow that political will, rather than economic constraint, might make the Fed slow to tighten (Section 6.1.4, p. 18).

Because most U.S. government debt is long-term rather than overnight, and long-term debt prices can absorb a shock to expected future surpluses in place of an immediate jump in the price level – and because the government can further choose to postpone inflation by selling additional long-term debt at the time of the shock, without any change in the underlying surplus path (Section 2.4 and 6.5, pp. 8-9, 21-23). Calibrating the U.S. debt maturity structure from the CRSP mbx database, Cochrane compares three inflation paths consistent with the same 10% shock to the present value of surpluses: an immediate one-time price jump, a steady inflation starting immediately, and a “postponed” inflation delayed four years by additional long-term debt sales, with correspondingly larger eventual inflation the longer it is delayed (Fig. 7, pp. 22-23). He concludes the plausible scenario involves “a puzzling rise in long-term interest rates” well before any actual inflation appears, with “steady inflation follow[ing] that, on a time scale roughly coincident with the average maturity of government debt” (p. 23).

He argues that off-balance-sheet nominal commitments (credit guarantees, defined-benefit pensions, nominally sticky government salaries) and growth effects on the present value of future tax revenue both mean the true fiscal limit is much closer than raw current-year debt/GDP figures suggest (Sections 6.3-6.4, pp. 20-21). On guarantees: government backing of Fannie/Freddie debt, TARP institutions, and (implicitly) “too big to fail” institutions functions as additional nominal debt, so that “having to make good on these guarantees on top of large budget deficits can be the piece of bad news that kicks expectations over the fiscal limit” (p. 20); he cites Novy-Marx and Rauh’s (2009) estimate of $3.23 trillion in underfunded state pension obligations against only $0.94 trillion in states’ publicly traded debt. On the “dynamic Laffer curve”: because what matters for the valuation equation is the present value of future tax revenue, and the present value’s growth term is divided by (r - g), “small growth effects can have a big impact on the fiscal limit” – Cochrane shows that with r - g = 0.02, a growth-rate reduction of just 0.02 (offsetting a 15% tax-rate increase from 30% to 35%) is already enough to put the government at its fiscal limit, disregarding any static output effect entirely (p. 21).

He argues it is far more likely to bring stagnation than a boom, because the fiscal valuation equation itself functions as the “anchor” for inflation expectations, so that a fiscal inflation operates as a Phillips-curve shift rather than a movement along a fixed Phillips curve (Section 7, pp. 24-26). Using a standard New-Keynesian model (output, a forward-looking Phillips curve, and a Fisher equation) and feeding in an inflation path similar to his “postponed inflation” scenario, Cochrane shows output declining throughout the entire episode: “This is stagflation, not a boom; a march of the Phillips curve up and to the right as in the 1970s. The reason is transparent: The inflation is all expected; expected inflation rises before actual inflation” (p. 25). He backs this with historical experience – 1970s stagflation, postwar hyperinflations, and Latin American and Zimbabwean currency collapses – arguing that “not all inflations come with output booms either in theory or in practical experience” (p. 24), directly against commentators (he names Mankiw and Rogoff) who favored deliberate inflation on the expectation that it would raise output.

Cochrane argues Fed statements throughout this period attributed low inflation to economic “slack” and treated inflation expectations as “stable” largely because inflation had recently been low (an adaptive-expectations argument), with no acknowledgment that fiscal sustainability, not just resource utilization, might be what is actually anchoring those expectations (Section 7.3, pp. 25-27). He quotes Chairman Bernanke’s March 2010 testimony attributing continued low inflation to “substantial resource slack” and “stable” longer-term expectations, and observes that “neither surveys nor long-term yields gave any warning of inflation in the 1970s nor disinflation in the 1980s” (p. 26). While noting that Bernanke did become increasingly vocal about the need for fiscal reform over 2009-2010, Cochrane argues the Fed’s own diagnosis of the risk channel – higher long-term rates and reduced “confidence” from a crowding-out mechanism – misses “the fiscal equation’s warning – that when investors question fiscal sustainability, inflation can break out despite ample ‘slack’ and there is nothing the Fed can do about it” (p. 27).

He concludes the Fed and the government as a whole may have very little power over either outcome under current arrangements: legally, the Fed can only ever exchange one government liability for another, so once any special liquidity premium on government debt is satisfied, its tools “run out of steam,” and it cannot by law undertake a genuinely fiscal action such as a “helicopter drop” on its own (Sections 5.1.7 and 8, pp. 18-19, 26-28). He frames this as a deliberate institutional tradeoff: “The rule roughly forbidding the Fed from direct fiscal action is the key to its independence. But as a result, the institution charged with maintaining the price level is forsworn from affecting the first-order causes of inflation!” (p. 19). Looking forward, Cochrane suggests the underlying instability stems from the government having no good mechanism to commit to and communicate a stable present value of surpluses, and floats financial-engineering alternatives – such as the Fed targeting CPI futures or the TIPS-Treasury spread directly, functioning like a modern commodity standard without requiring literal commodity purchases – plus greater reliance on long-term rather than short-term debt issuance, so that future surplus shocks show up as interest-rate variation rather than abrupt price-level or rollover crises (Section 8, pp. 27-28).