Dynamic stochastic general equilibrium (hereafter DSGE) models have been used as powerful tools for business cycle analyses. During the 1980s, 1990s, and 2000s, business cycle research is exploratory and researchers investigate effects of many different shocks, mechanisms that propagate them, and welfare implications in a consistent way that made clear what factors are important and why. There are many interesting fields in DSGE, in this dissertation, we try to investigate three of them : (i) news-driven business cycles, (ii) business cycle accounting, and (iii) short-run effects of technology shocks.

1 News-Driven Business Cycles

DSGE is often criticized on the ground that it does not provide a theory of recessions. It is well known that standard real business cycles models have difficulties explaining recessions without technological regress. When agents receive positive news (or have optimistic expectations) about the future, they decide to build up capital since future's aggregate demand increases. If the news turns out to be false, there will be a period of retrenched investment which is likely to cause a recession. Such effects of "news shock" might be one of important sources of business cycle fluctuations. A news-driven business cycle (hereafter NDBC) is a business cycle in which a positive news about future causes a current boom defined as simultaneous increases in consumption, labor, investment and output.

There are two major reasons why NDBC is highlighted in modern macroeconomics. One comes from empirical episodes. The Internet bubble of the U.S. economy during the late 1990s and the Japanese bubble era during the late 1980s might be accounted for by NDBCs; positive news about the future might cause such booms. The other comes from the theoretical side. It is well known that standard real business cycle (hereafter RBC) models do not generate NDBCs. News about the future moves consumption and labor in opposite directions due to the wealth effect in a standard RBC model. Therefore, one of the important challenges in macroeconomic theory is investigating what kinds of features should be introduced in a standard model in order to generate NDBCs.

In Chapter 2, we find that a fairly popular market friction, nominal rigidity, can be a source of NDBCs and that they can be generated by changes in markups in response to news about the future. NDBCs occur due to news about technology expansionary monetary policy shock in our model. We also find that the economy might fall into recession if news turns out to be false. The key mechanism of booms and recessions is that markups are moved by news through nominal rigidities. When the good news arrives, people expect that both inflation increase in the future, implying that the current optimal price level increases. But price-setting firms cannot fully increase their prices in response to the increasing inflation because of nominal rigidities, which is modeled as a Calvo-type sticky price and it leads to decrease of their markups. The decrease of markups induces an increase of aggregate demands. Then, output and labor input increase. Finally, household income becomes so high that both consumption and investment increase. If the news turns out to be false, the optimal price level decreases, but price-setters cannot fully decrease their prices because of nominal rigidities. This means the increase of markup, and the economy fall into recession. Our model also generates procyclical movements of Tobin's q (asset price).

In Chapter 3, we show that if the working capital, such as labor payment, is subject to the collateral constraint, and there are adjustment costs of investment, neoclassical business cycle models can generate NDBCs. The news about future technology growth and technology level cause NDBCs and the intuitive mechanism as follows. The news raises the price of capital today, which relaxes the collateral constraint. Since the wage payment is collateral constrained, the relaxation of the collateral constraint reduces the inefficiency in the labor market. It shifts the labor demand curve outward. If this force is sufficiently strong, it offsets the wealth effect on the labor supply schedule, and the equilibrium labor supply increases. So do output and investment. Consumption increases because of the wealth effect due to good news. We consider a models with collateralized capital, adjustment costs of investment, and heterogeneous of consumers: households and entrepreneurs. Our model generates NDBCs and procyclical movements of asset prices. Our model also generates a recession, if the news turns out to be false. This is due to heterogeneity of consumers; when the good news arrives, the price of the collateral asset increases, and hence entrepreneurs need less capital to achieve the desired value of collateral. Hence, in response to the good news about the future, entrepreneurs sell their capital. When the news turns out to be false, the price of capital essentially goes back to its steady state level, and it means too tightened collateral constraint. Then, a recession occurs.

2 Business Cycle Accounting

In the first generation of DSGE, technology shock is the main and dominant source of business cycle fluctuations. In the current generation of DSGE., many frictions and many shocks are proposed to account for the real economy. Then, researcher often faces hard choice about where to introduce frictions into model to construct model to generate business cycles similar to data. In other words, what types of frictions are promising to the research of business fluctuations?

Business cycle accounting (hereafter BCA) is a method to address this question. In BCA, the economy is assumed that it is described as a standard neoclassical prototype model with time-varying productivity, labor tax, investment tax, and government consumption. These are called efficiency, labor, investment, and government wedges. This assumption is justified by equivalence results; this prototype model with wedges covers a large class of frictional business cycle models. Wedges are measured so that the prototype model accounts for the observed data perfectly. Equivalence results are shown under general conditions about evolutions of wedges. However, in practice, they impose that wedges evolve according to the first order vector autoregressive, VAR(1), process and it is not clear whether conventional VAR(1) specification of wedges is consistent with conditions in terms of equivalence results.

In Chapter 4, we examine the equivalence results by focusing on the VAR(1)representation of wedges. We characterize the class of frictional models covered by the prototype model with the conventional VAR(1) specification of wedges. We find that the prototype model covers a detailed model if and only if wedges have sufficient information about the endogenous and exogenous states of the detailed model. Intuitively, the number of independent wedges should be larger than that of endogenous and exogenous states variables in the detailed model. We also show that the class covered by the prototype model is much smaller than that is shown under general conditions of wedges. Therefore, the condition for equivalence results is highly restrictive if we employ the VAR(1) representation of wedges. As a solution to this problem, we provide an alternative specification in order to let the prototype model cover much larger class and show that, in theoretically, it is possible to let the prototype model cover any class of frictional DSGE models by using our alternative specification.

3 Short-Run Effects of Technology Shocks

What are the short run effects of technology shocks? Economic theory provides two different predictions. Standard flexible price models like real business cycle theory predict that technology shocks are expansionary in the short-run; inputs and output increase immediately. On the other hand, standard sticky price models like New Keynesian's theory predict that technology shocks are contractionary in the short run; inputs and output decrease. Of course, both models predict that technology shocks are expansionary in the long run. The difference in the short-run is due to the flexibility of prices. It is important to address this question since the answer tells us the character of good economic model to draw the reality.

In Chapter 5, we use JIP (Japanese Industry Productivity) database, which is Japanese sectoral panel data, and identify technology improvements as purified TFP growth. Purified TFP growth is the one that the effects of (i) non-constant returns to scale of gross production function, (ii) variable utilizations of inputs, and (iii) sectoral reallocation, are excluded from the Solow residual. We focus on the relationship between technology shock and labor input in the short run as in related literature. We find that aggregate technology shocks are not labor-saving even in the short run and our result is robust to various cases. This is the first result which finds expansionary technology shocks for the Japanese economy by the Solow-Hall approach.

論文の内容

1980年代以来,少なくとも米・欧におけるマクロ経済学では,動学的確率的一般均衡モデル("DSGE (dynamic stochastic general equilibrium)モデルと通常略称される)を用いた景気循環や成長の分析が主流となっている。モデルが「確率的」と呼ばれる理由は,生産性の変化による供給ショックや,効用関数をシフトさせる需要ショックが,モデルの要素になっているからである。2004年にノーベル賞を受賞した Edward Prescott により,生産性ショックが景気循環を発生させることが解明されており,彼のモデルがこの分野の標準的なDSGEモデルとなっている。

奴田原氏の博士論文は,この分野での最新の二つの理論的トピックを扱っている。そのひとつは,将来の生産性についての新しい情報が引き起こす景気循環("news-driven business cycles",略してNDBC と呼ばれる),もう一つは,「景気循環会計」("business cycle accounting, BCA と略記)である。

NDBCについては,標準的DSGEモデルでは景気循環を発生させることが難しく,多数の追加的な仮定をもうけてモデルを複雑化する必要があると学界では考えられてきた。奴田原氏の貢献は,今や主流モデルとなった,価格硬直性がある標準的なDSGEモデルによってNDBCが発生することを示したことにある。これが論文の第2章である。

BCA(景気循環会計)では,標準的なDSGEモデルに,歪み("wedge" と呼ばれる)を導入することにより,DSGEモデルの集合を定義し,与えられたモデルがこの集合に属するか,属するとすればどのようなwedge を想定するのか,を分析する。この分析を行う場合,税率や政府支出といったモデルの外生変数について,その確率過程を想定しなくてはならない。BCAの提唱者は,確率過程として VAR(1)(1次の多変数自己回帰過程)を想定した。奴田原氏のBCAにおける貢献は,より一般的な確率過程を外生変数について想定すれば,いままでBCAの対象と考えられていなかった多くのモデルまで集合に含むことができることを示したことにある。この分析は論文の第3章に納められている。

以上が奴田原氏の博士論文に含まれた主要な理論的貢献である。最終章では,一転して生産性ショックについての実証分析をおこなっている。Prescott の標準的な景気循環モデルでは,生産性が上がれば,労働投入は上昇する。これに対し,ある種の価格硬直性が存在するDSGEモデルでは,労働投入は低下しえる。このように,生産性が労働投入に及ぼす影響を分析することにより,モデルの現実妥当性についての貴重な情報をえることができるため,1990年代以降,きわめて多くの論文が発表された。奴田原氏の貢献は,日本の産業別データを用いると,労働投入は増加することを発見したことにある。

評価

以上で要約したように,奴田原氏の博士論文は,NDBCとBCAについての理論分析と労働投入についての実証分析から成る。

まずNDBCについては,世界の第一級の学者が複雑なモデルで行ってきたことを,標準的なモデルに今や常識的となった仮定(価格硬直性)を課することで達成することができることを示した点で,この分野における重要な貢献となっている。ただ,複数の審査委員が指摘したように,NDBCが,その提唱者が主張するように日本の90年前後のバブルや米国のインターネットバブルの説明になるかについては,明らかでない。ただ,この点は,NDBC一般についてのコメントで,奴田原氏の貢献についての批判にはならない。

BCAについても,そもそもBCAがどの程度有用かについて,懐疑的な意見を持つ委員がいた。しかしBCAの有用性を認めた上であれば,奴田原氏の貢献は,きわめて明瞭である。論文では,明瞭な貢献をしているという点が必ずしもわかりやすくかけておらず,学術雑誌に投稿する前に,もう一回書き直す必要があるという意見もあった。

生産性が労働投入に及ぼす影響については,正の効果があるという結果は,日本のデータを使った研究では最初の発見であり,この分野の文献の貢献になるという意見が委員の多数を占めた。

最後に,論文のうち第2章(NDBCの分析)と第3章(BCA)は,共著論文であることについて,奴田原氏の貢献はどの程度なのかをいぶかる意見があった。この点については,主査である林が,すでに何回も奴田原氏がこれら二つの論文を発表し,林を含む複数の聴衆者からの難しい質問を奴田原氏が対応していたことを目撃している。奴田原氏がいわゆる「ただ乗り」をしているとは考えられない。

論文審査の結論

奴田原氏の博士論文は3つの主要な章からなっているが,まだどれも学術雑誌に publish されていない。しかし,上記の評価から明らかなように,どの章も,それぞれの分野での明瞭な貢献を含んでおり,適当な学術雑誌に投稿すれば,受理される可能性は高い。博士論文の少なくとも一章がすでに publish されているという条件は(もし今まで課されていたとすれば)課すべきでない。学界への有意な貢献を含む章が複数あるという意味で,奴田原氏の博士論文は,本研究科が要求する論文博士の基準を十分に満たしている。したがって,この審査委員会は,本論文により博士(経済学)の学位を授与するにふさわしいと全員一致で判断した。

2008年10月14日