Background

The Great Recession began in December 2007 and ended in June 2009, according to the U.S.

National Bureau of Economic Research (NBER). The emergence of sub-prime loan losses in 2007 triggered the recession and exposed other risky loans and over-inflated asset prices. With loan losses mounting and the collapse of Lehman Brothers on September 15, 2008, a major panic broke out on the inter-bank loan market. In the recession, the financial crisis played a significant role in the failure of key businesses, declines in consumer wealth estimated in trillions of U.S. dollars, and a downturn in economic activity leading to the 2008-2012 global recession.

The central debate about the origin of the recession has been focused on the respective parts played by the monetary policy and by the practices of private financial institutions. In order to strengthen the financial sector, the Troubled Asset Relief Program (TARP), in which assets and equity are purchased from financial institutions by the U.S. government, was enforced and originally authorized expenditures of $700 billion in October 2008.

Thus, it would be widely recognized and become a qualitative consensus that the solvency and liquidity problems of the financial intermediaries had a key role in the Great Recession. Then, fiscal and monetary authorities responded to this recession by injecting a large amount of public fund into the banking sector to improve the balance sheet. But, how can we quantitatively extract the impact of the bank sector’s balance sheet loss on the recession? Even if we measure the quantitative effect, was the deterioration of the balance sheet of the banking sector the main source of the recession?

Why does the negative impact of the banking sector expand so much? How should we quantitatively measure and quantify the policy effect on the recession?

How was the source identified in the past recession that the U.S. experienced and what was the method to measure the impact? Looking back on past economic turning points, the U.S.

131

economy had experienced the Great Inflation in the mid 1960s to the late 1970s (pre-Volcker era) and the subsequent Great Moderation since the mid 1980s (post-Volcker era). In the context of macroeconomics, there has been a long debate on what was the source of the Great Moderation.

There are two major arguments: One argument insists on “good policy” that the economic stability came as a result of boldly fighting inflation led by the Chairman Volcker at the beginning of the 1980s (Clarida et al. 2000, Bernanke and Boivin 2003), and the other argument claims simply shocks volatilities increased in the Great Inflation (“bad luck”) and decreased in the Great Moderation (“good luck”) (Cogley and Sargent 2005, Sims and Zha 2006, SW 2007, Justiniano and Primiceri 2008).

Clarida et al. (2000) insisted the policy response to inflation was rising during the Volcker era, which contributed to the Great Moderation in the 1980s, estimating monetary policy rule by GMM and examining the change of responses to inflation gap and output gap (changes in the Taylor coefficients). Bernanke and Boivin (2003) also reported that the Taylor coefficient for inflation was high during the Volcker era, estimating monetary policy rule using predicted CPI by DFM’s factor.

On the other hand, Sims and Zha (2006) concluded the main source of the Great Moderation is the decline of shocks volatilities since the 1980s. They estimated a structural VAR model allowing both regime switches of model parameters and shocks volatilities. Thus, they considered both possibilities that model parameters have changed (changes in monetary policy rules: good policy) and shocks volatilities have changed (good luck). As a result, they showed that the latter model was higher fit for the data.

In recent years, the sources of the Great Moderation have been examined using the empirical DSGE model, instead of reduced-form time series models such as DFM or structural VAR model.

SW (2007) estimated the DSGE model by dividing the observation period into two terms (Great Inflation: 1966:Q2-1979:Q2, Great Moderation: 1984:Q1-2004:Q4). There are three main findings:

First, the shocks volatilities declined in the second period, in particular, reducing volatilities of productivity shock, monetary policy shock and price markup shock. Second, (surprisingly) there was not much changes in both terms regarding steady state value of inflation and parameters of monetary policy rules. Third, the NKPC was flattened out in the second period through the rise in price and wage nominal rigidities (the rise of Calvo parameters).

To further clarify the sources of the Great Moderation, SW (2007) reports three counterfactual simulations: First, calculating GDP and inflation data volatilities in the second period using shocks volatilities estimated in the first period, then extremely volatile data can be reproduced. Second, calculating GDP and inflation data volatilities in the second period using parameters of monetary policy rules estimated in the first period, then the calculated volatilities did not change much.

Finally, calculating GDP and inflation data volatilities in the second period using all estimated parameters in the first period, then there also was not change much. From the results of the three experiments, the biggest reason why volatilities in the second period declined was not because the attitude of the monetary authority to inflation and GDP has changed (not a good policy) but because shocks volatilities have declined (but a good luck).

Instead of dividing the observation period, empirical analysis by explicitly incorporating time-varying shocks volatilities (or stochastic volatility; SV) has also been carried out. Thus, this ap-proach is to estimate the SV model using all of the observation data and examine the transition of the estimated time-varying volatilities. Justiniano and Primiceri (2008) argued the decline in the volatility of investment specific technology shock is the main source of the Great Moderation, by introducing SV to the SW model. Liu et al. (2011) reported the fall of the demand shock volatility called financial shock (capital depreciation shock) is an important source of the Great Moderation,

4.1. INTRODUCTION 133 by estimating the SW model that allows regime-switches for shocks volatilities.

However, the SW model has a problem that the friction in the financial market is not explicitly taken into account such as imperfections in capital goods market or bond market (asymmetric information on capital goods or bond transactions), or incompleteness of bond market (there is no market trading for state contingent claims, etc.). Nevertheless, an ad hoc shock, called as the equity premium shock, has been added in the transition equation of capital goods shadow price (Tobin’s q). As with markup shocks, it is impossible to find a structural interpretation of asset price fluctuations in such ad hoc shocks.

The Great Recession is triggered by the financial crisis as symbolized by the collapse of Lehman Brothers in September 2008. In recent years, empirical studies have been reported to extend to the DSGE model explicitly incorporating the financial frictions to examine the source of the Great Recession.

Carlstrom and Fuerst (1997) and Bernanke et al. (1999, BGG) are pioneering theoretical studies that introduced financial friction into the general equilibrium framework. BGG constructed a financial friction model resulting from asymmetric information between banks and firms on the gross return rate of investment projects.

In this model, firms invest in physical capital by receiving loans from banks, and the gross return rate of investment randomly realizes, but the gross return rate is private information for firms: It is observable for firms but unobservable for banks. In this case, firms have an incentive trying to reduce repayment by telling lies that the realized gross return on investment was low. To eliminate this incentive, banks will lend funds at a high rate (spread is added to the risk-free rate) according to firms’ balance sheet. In equilibrium, the firms’ borrowing rate is determined according to the debt ratio (or the leverage ratio) of the firms’ balance sheet and the spread has been shown to be an increasing function of firm’s leverage ratio.

Under such circumstances, a slight decline in firm’s own asset price or a minor damage of capital stock quality will increase the leverage ratio, raises the spread, and the borrowing rate rises, resulting in firms facing high borrowing constraints. As a result, investment will decrease. In addition, the rise in the borrowing rate lowers the net return rate on investment (= gross return rate on investment - borrowing rate). As asset price is the discounted present value of net returns of future investments, declines in net returns will result in a further decline in asset price. A decline in asset price further increases leverage ratio, which will raise the spread and the borrowing rate will rise. As a result, further investment and net return will decline.

In this way, a trivial negative financial shock, such as a decline in asset price will result in a substantial decline in investment by an amplification effect called as a financial accelerator mecha-nism.

On the other hand, Gertler and Kiyotaki (2010, hereinafter, GK) and Gertler and Karadi (2010) propose a DSGE model introducing a friction into financial transactions between banks and depos-itors. In this model, depositors are lenders and banks are borrowers, and the financial friction is caused by bank’s moral hazard behavior that banks can convert deposits into their assets. In this case, an incentive constraint should be imposed by depositors so as for bankers to continue the banking business. As a result, depending on the bank’s net worth ratio (= 1 - debt ratio), depositors will discipline banks through the deposit amount. After all, if bank’s net worth ratio declines, spread between deposit rate and risk-free rate will rise, banks will face higher borrowing constraints, and investment will be hampered, similar to BGG, the financial accelerator mechanism will be generated.

Turning to the empirical side, Christensen and Dib (2008) is one of the earliest studies that

estimated the DSGE model with financial friction. They estimated models incorporating BGG type financial friction and without financial friction, using the data during the Great Moderation in the U.S. As a result, they reported the data fits better for models incorporating financial friction.

Kaihatsu and Kurozumi (2014a, b) explored the sources of the Great Recession in the U.S. and the determinants of the lost decade in Japan, similar to Christensen and Dib (2008), by estimating a model with the BGG type financial friction. According to the result, an adverse financial shock (declining asset price) is the main source of the Great Recession in the U.S., but in Japan, as in Hayashi and Prescott (2002), a contraction supply shock (negative investment specific technology shock) was the main factor of the recession.

Recall that BGG is the financial friction model between banks and corporates, and the friction is based on private information of corporate sector, so the incentive constraint is imposed to the corporate sectors’ balance sheet.

Since the recent Great Recession of the U.S. and Japan’s lost decade, however, caused by the collapse of financial intermediaries such as investment banks, commercial banks, securities companies, etc., there is a possibility that the main source was the deterioration of the bank’s balance sheet rather than damaging the corporate balance sheet.

If so, how should we know which deterioration of the balance sheet of the banking sector and the corporate sector was the main factor behind the Great Recession?

To detect the origin of the recession, we need to decompose the economic downturn into two adverse financial shocks. Hirakata et al. (2011) expanded to a model introducing two financial frictions, i.e. one is the friction between banks and corporates and the other is the friction between depositors and banks. The banking sector mediates the supply of funds from depositors to the corporate sector. Thus, bankers are lenders of firms and borrowers from depositors. They introduced agency costs due to asymmetric information, i.e. BGG-type financial friction, in both banking sector financial transactions. Then, they demonstrated the banking sector’s net worth deterioration was a major factor in the substantial decline in investment.