Breaking down the Blinder/Zandi paper (Part 2)

Click here for a printer-friendly version of this article

As noted in a previous post, the Blinder/Zandi paper concluded that the TARP stabilization measures combined with the Fed’s quantitative easing had a greater positive effect on subsequent economic growth than the Obama stimulus.  It is important that these sets of policies be treated separately when quantifying the effect of overall government intervention.  There is a profound difference between policies designed to prevent a collapse and those designed to speed the pace of recovery.  Viewing these policies as parts of the same “government intervention” trivializes distinctions that are central to the lessons policymakers should draw from the response to the crisis.

The financial crisis of 2008 was like nothing the economy had experienced since the Great Depression.  Lehman Brothers’ unsecured creditors lost nearly everything in the bankruptcy filing (the value of senior bonds fell to 8.75 cents on the dollar), clients’ accounts in London were frozen, counterparties used powers granted under bankruptcy law to seize collateral, and general panic ensued after the Reserve Fund (an unsecured creditor to Lehman) failed.  After Lehman, Wachovia was sold, Washington Mutual failed, Goldman Sachs and Morgan Stanley converted to bank holding companies, and virtually every financial firm that relied on short-term borrowing to finance illiquid positions in securitized assets was on the brink of failure.  

Had the government not intervened at this point, financial intermediation in America – and the world – could have very realistically stopped functioning.  As Bernanke (1983) showed, the collapse of the banking system was a major driver of the fall in output witnessed between 1929 and 1933.  The disruptions of this period prevented the banking system from performing its basic economic functions, which compromised the ability of borrowers to access credit to finance investment and consumption.  It was only after the banking system had been stabilized through the bank holiday and recapitalization of 1933 that the economic decline was finally arrested.

Without the actions taken in the fall of 2008, including the Fed’s purchases of commercial paper, mortgage assets, government bonds, and acceptance of a wide range of securities as eligible collateral for loans, the economy might have endured a four-year contraction similar to 1929-1933.  It is not a surprise that this scenario is very close to the counterfactual baseline used in the Blinder/Zandi paper.  The problem is that once one concedes that a Depression was prevented by the federal government’s commitment to backstop financial intermediation, it becomes difficult to discern any positive economic effect from the subsequent stimulus bill.

The actions taken to stabilize the financial system dampened the amplitude of the shock, but could not stop its effects from rippling through the rest of the economy.  The impulse response to the shock was very large, with output and employment levels falling very rapidly.  But as with the seismic recordings following an earthquake (see chart), the waveforms generated by the shock eventually die out.  By the middle of 2009 the economy had largely recovered from the shock and returned to almost the identical path it was on in 2008 – slow growth, modest declines in payrolls, and mixed readings on whether the economy was or was not in what could technically be called a recession.

But the only effect of stimulus at this point had been modest reductions in withholding, with most of the spending to come later in 2009 and 2010 (and beyond).  The counterfactual “baseline scenario” for the effect of stimulus should be 2008’s slow growth, not the Great Depression-like depths that had already been taken off the table by the decisions made in late 2008.  

The study’s finding that the stimulus had a large positive effect on output is an example of how the use of Keynesian economic models in search of Keynesian economic effects prejudges the results.  As Arnold Kling writes, “every result in the paper would have been found by simulating the model three years ago.”  Moreover, there is no mechanism in such models to parameterize non-Keynesian (or supply-side) channels through which debt-financed stimulus impacts the economy.  The much talked-about decline in business confidence has actually been a common response to large increases in debt burdens in other nations.  By creating expectations of distortionary tax increases, debt-financed stimulus can reduce permanent income, which reduces household consumption and business investment. 

As the unemployment level remains stubbornly high, the Obama Administration’s rhetoric has focused on the depression that would have ensured in the absence of stimulus.  Critical to this effort is either ignoring the 2008 policy response or conflating it with stimulus into a single “government intervention.”  Close reading of the Blinder/Zandi paper makes clear that even the models most favorably disposed towards finding a big effect for stimulus show that the financial stabilization policies played a much bigger role.  

An analysis of the effect of stimulus that properly re-adjusts the baseline to 2008-like growth and accounts for non-Keynesian effects on private sector expenditure would likely find that the stimulus has had little net effect on output, or worse.