This item is only available as the following downloads:
BANK LIQUIDITY PREFERENCE SHOCK S AND MACROECONOMIC FLUCTUATIONS BY KATHLEEN MCQUEENEY A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Tarron Khemraj Sarasota, FL May 2012
ii Acknowledgements First and foremost, thank you to my parents for their unconditional love and support, especially throughout the past four years. They taught me that responsibility, ambition, and intellectual curiosity would get me far in life, and without their guidance I would not be the person I am today. They knew that New College would be a great fit for me far before I did, and even though I fought them all of way, they were right. Thank you to Tarron Khemraj, who has been my advisor and mentor at New College since my first day. I would like to thank him for helping me throughout the thesis process and giving me great advice about my life post New College. I would also like to thank other faculty members who have been particularly supportive in the past year includin g Patrick Van Horn and Duff Cooper for sitting on my committee, as well as Barbara Hicks and Rick Coe for helping me through the fellowship and graduate school admission process. It is important to note that I would be nowhere without the invaluable guidan ce of Elliott who has served as a role model for me since I met her my first year. She made me confident in my academic work and that has helped me through the thesis process despite her absence in my final year. A big thank you goes out to the New Colleg e community for being my home for the past four years. My involvement in student government and work in jobs on campus have given me opportunities that have enhanced my education more than I could have ever imagined. I would like to thank all of the amazin g friends I have made in the past four years you know who you are My New College memories are so much better with you in them.
iii Table of Contents Introduction ................................ ................................ ................................ .................... 1 2 Chapter 1: Literature Review I. Importance of Prior Literature and Old Keynesianism ................................ .... 3 6 II. Monetarism ................................ ................................ ................................ ..... 6 10 III. New Classical Theory ................................ ................................ .................. 10 12 IV. Real Business Cycle Theory ................................ ................................ ........ 12 15 V. New Keynesianism ................................ ................................ ....................... 15 20 Chapter 2: Significance of Bank Liquidity Preference I. Why does bank behavior matter? ................................ ............................... 21 2 5 II. The Influence of Financial Liberalization ................................ .................. 25 26 III. The Backdrop of the Recession ................................ ................................ 26 29 IV. Why are banks holding so many excess reserves? ................................ .... 30 34 V. The Importance of Bank Liquidity Preference ................................ .......... 34 3 6 Chapter 3: Methodology I. Model of Bank Demand for Reserves ................................ ........................ 3 7 4 4 II. Model of the Effect of Bank Liquidity Preference Shocks on Macroeconomic Indicators ................................ ................................ ................................ .... 44 47 III. What is the likely effect of liquidity preference shocks on macroeconomic indicators? ................................ ................................ ................................ 47 53 Chapter 4: Results and Discussion I. Effect of Bank Liquidity Preference Shocks on Macroeconomic Variables Using Long term Coefficients ................................ ................................ ... 54 64 II. Effect of Bank Liquidity Preference Shocks on Macroeconomic Vari ables Using Impact Coefficients ................................ ................................ ......... 64 67 III. Limitations of Work and Future Research ................................ ................. 6 7 69 Conclusion ................................ ................................ ................................ .................. 70 71
iv BANK LIQUIDITY PREFERENCE SHOCKS AND MACROECONOMIC FLUCTUATIONS Kathleen McQueeney New College of Florida, 2012 ABSTRACT In the years since the financial crisis in 2008, U.S. commercial banks have seen an unprecedented increase in liquidity preference. This thesis examines the relationship between bank liquidity preference and the business cycle. Shocks in bank liquidity preference are estimated and identified total reserves and the loan rate and the federal funds rate. The thesis notes that a well defined liquidity preference curve is evident in the credit market. Shifts in the curve represent shocks in bank liquidity preference The purpose of this thesis is to measure the extent to which both negative and positive ban k liquidity preference shocks affect unemployment, GDP, inflation, corporate profits after taxes, the stock market, and housing starts. The effect s of bank liquidity preference shocks on many of these macroeconomic indicators were found to be economically significant, suggesting the importance of this bank behavior in the business cycle literature. Dr. Tarron Khemraj Division of Social Sciences
1 Introduction Since 2008, the relationship between the banking sector and the economy at large has become more apparent than ever. The financial crisis sent shockwaves throughout the economy that have had a lasting effect, as can easily be seen the negative effects on m acroeconomic indicators. Obviously the financial crisis has affected the life of the average American through higher unemployment and a generally bleak economic outlook for the future. The drastic downturn in the business cycle has made the relationship be tween the financial sector and the overall economy especially important to consider. As the economy emerges from the recession at a sluggish pace, it is important to examine the mechanisms through which the financial crisis and specifically actions taken b y financial intermediaries affect the macroeconomy. The banking sector seems to have a fairly large effect on the general economy. Given this relationship, bank behavior should have a significant place in business cycle theory, which attempts to explain what drives fluctuations in the economy. Bank behavior seems to be especially important since the age of financial liberalization stripped many bank regulations and gave financial intermediaries a greater sense of autonomy. However, this is largely absent from the literature. One clear change in the behavior of commercial banks since the financial crisis has been an unprecedented growth in liquidity preference, as their total reserves have increased by a staggering amount. Instead of that money going into t he economy, it is held in banks, which seemingly would negatively impact the economy especially as money is not being used for lending purposes. Many reasons for holding this money are a result of the recession, wh ich has made the flow of money less trans parent. The recession has changed liquidity preference so drastically in part because of the uncertainty inherent during these types of economic
2 periods. The effect of bank liquidity preference on the economy is virtually nonexistent in the business cycle literature, so this thesis hopes to contribute to the literature by considering the relationship between bank liquidity preference and macroeconomic variables to give a sense of how this kind of bank behavior affects the broader economy. In order to exami ne this relationship, the thesis seeks to first identify liquidity preference shocks by banks and then estimate their effect on different macroeconomic variables. This is done in two general steps, the first of which estimates banks demand for reserves to find the shocks themselves. The second step assesses the relationship between the shocks in bank liquidity preference and changes in macroeconomic variables, which represent changes in the economy. Times series econometrics plays a key role in this thesis because the addition or removal of money from the economy takes time. When the relationships between bank liquidity preference shocks and key macro variables demonstrate statistical and economic significance, this indicates that shocks in bank liquidity preference do help drive fluctuations in these variables. The meaning of these relationships suggests that bank behavior in regards to liquidity preference does help drive changes in the business cycle. As banks continue to hold an unprecedented amount of reserves, this may give a sense of why it has taken the economy so long to come out of the Great Recession that hit in 2008.
3 Chapter 1: Literature Review Theories of the business cycle are particularly relevant today given the recent global recession. In the midst of the worst job economy since the Great Depression, economists continue to search for the reasons behind the slow progress of the economy since the financial crisis. Looking to past theories that attempt to explain why such downfalls in the economy happen are helpful in assessing the current situation, although none seem to be able to fully explain the trends that have been evident during this trough. Evaluating these perspectives reveals a lack of thorough institutional consideration, especi ally in regards to the interconnectedness of the financial sector and the broader economy. The relationship between the banks and the economy as a whole has become all the more evident since the financial crisis brought the entire American and global econo my down in 2008. The business cycle refers to fluctuations in the economy over time, specifically in terms of changes in output, which are usually measured by the growth rate of gross domestic product (GDP). Different approaches to business cycle theory attempt to explain long run growth trends typically between periods of economic expansion and contraction. While some approaches treat the causes for changes in the business cycle as exogenous and hold that the causes behind trends in the business cycle co me from external sources, others maintain that the business cycle is driven by endogenous factors, namely issues of demand. As modern macroeconomics was born from Keynes, the review of the business cycle literature will begin with Old Keynesianism, progres s through the ideas of monetarism and new classical economics, and conclude with the New Keynesianism to examine how past theories explain what drives recessions in
4 particular. It is clear that none of the theories are able to fully explain the state of th e stagnant nature of the current economy. I. Old Keynesianism Keynes revolutionized the field of economics during the Great Depression, the greatest trough in American economic history. He focused his studies on understanding what was driving the contraction in the economy, just as economists are doing today. Keynes changed the discipline of economics by focusing his analysis of the economy on output levels at specific points in time. This helped mark the beginning of the business cycle literature as economis ts began to view output as an important indicator of growth trends. He asserted that the business cycle is driven by positive and negative shocks to real aggregate demand and rejected the classical notion that the demand for money in an economy is stable ( Snowdon and Vane 2008 ). The emphasis on shifts in real aggregate demand causing fluctuations in the economy indicated that human behavior is an important factor when considering the forces behind economic trends because these behaviors prevent the market from being able to clear. This is largely because with a decrease in real aggregate demand comes an increase in unemployment and a decline in real output (Tobin 1993). Keynes argued that one of the main drivers of the Great Depression and its pervasive nat Great Depression believed that the economic future was bleak, especially in terms of unemployment, they tended to save more and spend less. This had negat ive ramifications in the economy in terms of demand, as many of these savings came in the form of cash kept personally instead of savings that could be converted by banks into investment.
5 During times like these, investment tends to remain low because bus inesses are unsure of the future as well. Feelings of uncertainty often lead to instability which can serve to undermine the economy as they did in the 1930s. These negative feelings help create the inconsistencies that are evident in the economy over time (Akerlof and Shiller it revealed that the demand for money is not constant, as classical economics had suggested. The demand for money, also known as liquidity pr eference, is affected by perception, and this relationship can have implications in the broader economy. Keynes introduced the idea that liquidity preference of individuals has the ability to expand and contract the economy through its affect on aggregate demand. From the hold the most liquid forms of money, like cash, instead of other less liquid assets. Individual liquidity preference is determined by income level, the interest rate, and feelings about where the economy is heading. Liquidity preference is important because as people demand more cash to save and not to spend, consumption and investment can decrease, contracting the economy overtime and negatively impacti ng full employment and potential output. The Old Keynesian liquidity preference function considers the amount of cash holdings in the economy against the interest rate. There are speculative and transactions motives for holding money. The transactions mo tive is the need to have relatively liquid money for day to day needs and the precautionary need for money in the case of an emergency. These motives are not particularly interest rate sensitive, unlike the speculative motive for holding money ( Snowdon and Vane 2008 ).
6 The speculative motive for holding money considers the interest rate as it serves as a reward for a being less liquid in a given period of time. It essentially determines when people hold cash instead of less liquid financial assets that are a vital part of investment. The speculative motive considers the sense of uncertainty that individuals may have towards the economy. This is important when considering what drives the business cycle and makes the interest rate somewhat dependent on the ge neral confidence of the economy. As people hold more cash during times of uncertainty, hurting the supply to entice borrowing and get the economy moving again. While moneta ry policy is important during recessionary periods, ultimately it becomes ineffective as the interest rate can only go so low. Additionally, liquidity preference is hard to change given the Snowdon and Va ne 2008 ). This demonstrates the need for significant government intervention via fiscal policy to reverse the downward trend of the economy during recessions. Ultimately much of the information from the Old Keynesians in terms of the business cycle has proven to be important during the most recent recession in terms of the lack of efficacy of monetary policy at a certain point and the strong power of animal spirits over a number of economic agents. The juxtaposition of cash holdings and the interest rate will serve as an important relationship later in the thesis. II. Monetarism Monetarists, led by Milton Friedman, disagreed with the Old Keynesian idea that investment is vital in understanding the fluctuations in the economy. Their evidence demonstrated tha t there was no real correlation between autonomous expenditures and
7 consumption for a given stock of money. Instead, the monetarists held that the stock of money, or money supply, displays a consistent relationship with changes in the economy as a whole an d thus is central to business cycle theory (Friedman and Schwartz 1963). Additionally, Friedman reinstated the classical notion that the demand for money is essentially neutral. Monetarists argue that the theory of demand for money is equivalent to the qua ntity theory of money, which is the idea that the price level and money supply have a proportional and direct relationship. According to this framework, the demand for money is a stable function of a few variables, including permanent income, the return on financial assets, the expected rate of inflation, and tastes and preferences ( Snowdon and Vane 2008 ). The neutrality of liquidity preference helped emphasize that the money supply was the most important variable in explaining changes in the economy. The monetarist analysis followed that, all other things constant, the demand for money will decrease the lower the income level, the higher the yield on other assets, and the higher the expected rate of inflation. Wealth will be reallocated between different f inancial assets when the marginal rates of return are not equal. The reallocation of wealth is key in the monetarist approach to the business cycle as changes in the money supply affect the real output and unemployment. These changes can be illustrated by looking at open market operations. An initial equilibrium between marginal rates of return on money, financial, and physical assets is assumed. When the Federal Reserve buys bonds, the money supply increases because the public holds more money. Subsequent ly the marginal rate of return on holding money decreases as there is more of it available in the economy. The most liquid forms of money are exchanged for financial and real assets, and as demand for
8 these products increases, so will their prices. This de monstrates the relationship between the money supply and the price level that Friedman emphasized in the close tie between the quantity theory of money and the quantity demand of money. Prices will increase until there is an equilibrium in the portfolio in terms of marginal rates of return. This framework demonstrates the monetarist idea that changes in the money supply drive the business cycle, as an increase in the money supply incentivizes the purchase of physical and financial assets, which encourages c apital investment, ultimately helping spur growth in the economy. Friedman used empirical evidence to support this theory, although his methods were often criticized. He attributed monetary growth to key turning points in U.S. economic activity. His evide nce demonstrated that growth in money supply preceded peaks in the economy and was slower during economic contractions. While in the Old Keynesian framework monetary changes were viewed as consequences of recession or expansion, monetarism views changes in the money supply as the major reason for changes in the money income of the U.S. As income rises, the demand for money increases proportionately, as permanent income is a variable in the liquidity preference function. As liquidity preference of the public is viewed as stable in this sense, most instability in the economy is due to fluctuations in the money supply caused by monetary authorities. The monetarist policy conclusion is to allow the money supply to grow at a fixed rate along with the underlying g rowth of output to ensure long term price stability ( Snowdon and Vane 2008 ). augmented Phillips curve using expected inflation. Using the money supply as a causal
9 link, monetary expansion increases aggregate demand, which reduces unemployment below its natural level. At this point, the expected inflation rate is zero. Upward pressure on prices and wages results from excess demand for goods and service. As expected inflation was ze ro and price stability was evident, workers see the increase in nominal wages as an increase in real wages and subsequently supply more labor. Real wages fall, as does unemployment as firms demand more labor. Nominal prices and wages rise and unemployment falls, and this negative relationship is demonstrated by the downward slope of the short run Phillips curve. However, once inflation expectations begin to change and workers realize their real wages have actually fallen, they urge for raises. Workers as i ndividuals have incomplete information which ultimately causes fluctuations in the economy. This leads to a shift in the short run Phillips curve to the right, as wages rise to account for the increase in inflation. Firms lay off workers as real wages incr ease and unemployment increases until unemployment readjusts to its natural level. As the rate of inflation becomes fully anticipated, there is no long run tradeoff between unemployment and wage inflation in wage bargaining, and there will be a long run Ph illips curve at the natural rate of unemployment ( Snowdon and Vane 2008 ). More broadly, the increase in money growth indicates that the Fed has increased the money supply, which ultimately reduces the nominal interest rate and the real interest rate for a given level of expected inflation. This increases output along the aggregate expenditure curve, thus decreasing unemployment and increasing inflation along the Phillips curve. Ultimately the focus on the money supply as a driver of the business cycle seem s problematic, as the Fed has attempted to spur economic growth in the economy through
10 increasing the money supply to decrease the interest rate, and it has not worked well enough. It seems that the efficacy of growing the money supply to drive the economy out of its current trough has hit a plateau in the sense that there are diminishing returns to the increased amount of money being pumped into the economy by the Fed, suggesting a liquidity trap. Additionally, the emphasis on expected inflation is unfound ed today given that the increase in the money supply has not led to a big increase in inflation, meaning that workers asking for a raise in the midst of inflationary expectations are not spurring the persistent unemployment problem. III. New Classical Theory The new classical school took an equilibrium approach to modeling the business cycle and concluded that all business cycles are inherently alike. In this perspective business cycles are considered fluctuations around an overall real output trend that is c orrelated with changes in the price level, consumption, profits, investment, the money supply, productivity, and the interest rate. While changes in the money supply affect output in the short run, there is a neutrality of money in the long run so money gr owth affects price but not output. In this sense, the Phillips curve is a central feature of the new classical approach to the business cycle, given that the long run Phillips curve is vertical at the natural rate of unemployment at all levels of inflation This demonstrates the prominence of the general equilibrium. Another important aspect of this framework is the idea of rational expectations, which maintains that individuals process and utilize information to forecast the future in a way that fosters st ability. Thus, instability in the business cycle occurs because of imperfect information ( Snowdon and Vane 2008 ). The new classical model of the business cycle holds that unanticipated changes in the money supply lead to unexpected shocks to aggregate dem and which affects the
11 whole economy. This causes errors in price expectations and results in deviations in the long run equilibrium levels of output and unemployment, based on perfect information. Workers and firms mistake nominal price changes for real on es, and this consequently changes the supply of both labor and output. Business cycles are driven by exogenous monetary demand shocks that spread imperfect price signals to agents in the economy. In a world of imperfect competition, these economic agents r espond to price increases by increasing supply. In the wake of a monetary shock, greater general price variability yields a lesser fluctuation in output. This is different from the monetarist perspective in the sense that asymmetric information is evident on the part of both the individuals and the firms, leading to fluctuations in the economy. It is also different because the demand for money is not considered neutral and is seen to have a significant effect on the economy. The new classical model begins at an intersection of the aggregate demand curve (AD), the short run aggregate supply curve (SRAS), and the long run aggregate supply curve (LRAS). At this point, the expected and actual price levels are equal and output and employment are at their natural, long run levels. If the Fed decides to increase the money supply, this likely leads to an increase in the price level as aggregate demand increases and AD shifts outwards, necessitating a demand for more money. The chain of events also means that the actual price level is now lower than the expected price level. However, this information is not perfectly reflected in the economy, as it takes time to adjust, and often supply will increase in reaction to the increase in the pr ice level. Rational economic agents with full information will take the price increase into account and supply will actually decrease, as wages increase due to the increase in the price level.
12 This means that SRAS will shift inwards and will come back in l ine with LRAS, or the natural rate of unemployment ( Snowdon and Vane 2008 ). This perspective holds that if global prices increase nominally, output and employment will go above their equilibrium levels. Once they realize there has been no change in real pr ices, output and employment fall back to their natural levels, even though this would necessitate a contraction of the economy. Thus, new classicists believe that changes in the money supply, or monetary policy, have no effect on output in the long run an d only affects the price level. Monetary policy can cause fluctuations in the economy in the short run, but in the long run the economy is always trending back towards a single level of output and unemployment. In large part these problems associated with monetary policy have been evident recently. In this sense, all economic fluctuations are alike. However, the new classicist rationale has had some major issues in light of the most recent recession, as it has been unlike any other recent troughs given the large rate of unemployment and major dip in output. The latest episode of the business cycle is not like the others, and this is largely because of the influence of the banking sector. The rational expectations aspect of this school of thought is also pro blematic, as individuals do not have perfect information and neither does the market. This has been made evident by the inability of the market to predict the financial crisis and properly price a number of assets. IV. Real Business Cycle Theory Real business cycle theory assumes that economic fluctuations are due to large random technological shocks. These supply side shocks lead to fluctuations in output and employment as individuals change both their labor supply and consumption practices,
13 given rational ex pectations and changes in the relative price structure in the economy. These include natural disasters, political unrest, energy price shocks, new government regulations, and changes in productivity. Productivity or technological shocks are most significan t because they have the ability to affect in the economy in the long run. Modern real business cycle proponents disagreed with the idea that monetary forces drive the economy and attempted to discredit models that rely on fluctuations in aggregate demand t o explain instability ( Snowdon and Vane 2008 ). This marks a serious departure from the previously mentioned theories. Real business cycle theory came to prominence in the 1970s in part due to large increases in the price of oil which emphasized the destab ilizing abilities of supply side factors in terms of the economy at large. The inability of Old Keynesianism to explain increases in unemployment and increasing in inflation also allowed an entirely different point of view to gain reputation. Real business cycle theory became an integral part of the neoclassical movement of economics in the 1980s, which also focused on the Efficient Market Hypothesis, incentive based economic growth, and the importance of government failure. Real business cycle theory is r elated to new classical theory in a number of ways, however. Technological shocks work through the economy via impulse and propagation mechanisms. The impulse mechanism is the first part of the shock that causes a variable to veer from its steady state. Th e propagation is made up of the forces that keep the shock going forward in time and make the deviation from the steady state something that persists. Economic agents maximize utility in light of limited resources and have rational expectations. They have the ability to recognize if shocks are temporary or long lasting.
14 In this model, price flexibility is emphasized, not the price stickiness that is a staple of the Keynesian perspective, and ensures that markets clear. Additionally, propagation mechanisms like consumption smoothing and investment lags exist to carry shocks out in the economy. Changes in unemployment are purely voluntary, in the sense that workers change the number of hours they choose to work and leisure and work are substitutes. Ultimately money is neutral, and subsequently monetary policy has no effect on the economy, even when it comes to prices. Finally, there is no difference between the short and long run ( Snowdon and Vane 2008 ). This indicates that imperfect information is no longer v iewed to be important when it comes to changes in the economy, meaning cycle. Real business cycle theory holds that output accounts for consumption and investment an d is a function of capital and labor. This function is amplified by a total factor productivity level, also known as a technology shock. The technology level in a given period depends on the level in the previous period and the residual, which is a disturb ance. All individuals are considered to be identical, meaning that individual utility functions are able to be representative of the broader economy. Utility is a function of consumption, which is possible because the individual works, and leisure. In this sense, consumption (or work) and leisure are substitutes. As the individual will maximize utility over time, consumption and hours of work are taken into account. Limited resources come into the equation through the capital function, which consists of sav ings and depreciation variables. Savings is equivalent to investment, as all savings are intermediated by banks to be invested in the economy. A technological shock will cause a
15 response by the utility maximizing economic agent, leading to changes in consu mption, hours worked, investment, and thus output ( Snowdon and Vane 2008 ). This model is highly stylized, and the emphasis on perfect information and expectations reverberated through the neoclassical takeover of economic theory. While it is true that the supply side was largely ignored by new classicists and Old Keynesians, the lack of regard for the demand side in this view is also problematic. Ultimately, the theory was based on exogenous technological shocks causing fluctuations in the economy. Real bu siness cycle theorists essentially believe that demand shocks only affect the economy when they are brought on by changes in government expenditures and preferences. These ideas created an extremely anti government regulation sentiment in economic theory a nd especially when it came to mitigating the volatility of the business cycle. This has been an issue lately, as has the argument that unemployment is completely involuntary. Additionally, the market has not been able to clear and bring the economy back to its equilibrium. While monetary policy has not proven to be extremely potent, it has had some positive effects in the recent trough. Most importantly, there was no major technological shock that led into this recession, showing that this framework is unab le to explain different swings in the business cycle. V. New Keynesianism The parallels between the Great Depression and Great Recession have allowed Keynesianism to come back to a point of prominence. Even before that, Keynesianism began resurging as new t heory was added to make the Keynesian explanation of business cycles more complete. A new Keynesianism rose to prominence, which focused more on modeling the macro economy with microeconomic foundations while using general
16 equilibrium. Imperfect informatio n continued to be embraced by the New Keynesians as a reason for why fluctuations in the economy occur. Ultimately, New Keynesians made strides to go beyond the demand shock framework of the business cycle to include supply side variations and their effect on the economy as a whole ( Snowdon and Vane 2008 ). The New Keynesian school is a varied one, and its ideologies are not particularly streamlined. There is no unified view on government intervention, especially when it comes to fiscal policy. The stabiliz ing role of monetary policy is much more important in the New Keynesian tradition, but the willingness for the government to take an activist approach is more contentious. However, the hard and fast money growth rate rule proposed by Friedman is dismissed by this framework. Additionally, most New Keynesians still believe in rational expectations, which is the idea that people will do whatever maximizes their utility given all available information. This is an idea that has faced criticism more recently. At the core of the New Keynesian school are the ideas that money is not neutral and imperfect information is the key to understanding the business cycle. This is framed in terms of prices, as the stickiness of prices prevents money from being neutral, and th e explanation for the price rigidity is imperfect information. New Keynesians believe that both demand and supply shocks are relevant in understanding fluctuations in the economy but do not necessarily believe that the market has the ability to fully absor b a shock that will allow it to go back to the general equilibrium without government intervention. This is because both demand and supply shocks are amplified to a large extent by imperfect information and price rigidities. In this sense, involuntary
17 unem ployment is likely to exist in the economy ( Snowdon and Vane 2008 ). This view seems more practical given the amount of unwanted unemployment today. Within the new Keynesian framework there have been two sub groups that examine economic fluctuations, one t hat emphasizes nominal price rigidities, and another that focuses on the negative impact of price flexibility. The first is more popular and is embraced by those who looked for a more viable explanation of what really happens in the economy to provide an a lternative to the Real Business Cycle view (Mankiw 1989). The latter is akin to the original approach adopted by Keynes and Tobin. The first point of view sees the impact of a shock in aggregate demand as follows. Consider a decrease in the money supply t hat shifts aggregate demand inward. Real price rigidities, or menu costs, keep the price level constant at a particular level. This reduces output in the economy to a lower level, ultimately reducing the demand for labor. While this moves the level of labo r demand off of the demand curve, there is no market for the extra output necessary to hire an equilibrium level of workers. The negative aggregate demand shock causes an increase in involuntary unemployment, as the price level remains fixed and the short run aggregate supply curve is perfectly elastic. After a potentially unacceptable amount of time, downward pressure on prices and wages would move the economy back towards the long run aggregate supply equilibrium, allowing the economy to come back to a ge neral equilibrium. The New Keynesian model advocates measures to move the economy back towards its original equilibrium instead of just waiting on the market to clear. Usually the measures needed to move the economy back towards potential output and full e mployment include government interventionist policies. The level of intervention necessary to do, however, is a hotly debated subject.
18 The inability of the economy to adjust prices downwards more quickly is the result of a coordination failure. A coordin ation failure happens when there is no incentive for one economic agent or firm to adjust down prices assuming that other agents will not do the same ( Snowdon and Vane 2008 ). This demonstrates some of the microeconomic foundations of the New Keynesian mode l. The second view in the New Keynesian framework suggests that the main problem in terms of economic fluctuations is not price rigidity. This view holds that even if prices were inflexible, there would still be instability in both output and unemploymen t. Firms must balance risk aversion and a sense of uncertainty in the broader economy, leading to fluctuations in the economy. Asymmetric information causes imperfect information issues in financial markets that can restrict equity finance for firms. Thus, they must rely on debt and not equity finance, making bankruptcy much more likely in times of recession. It is easier for firms to reduce output instead of prices because of the uncertainties associated with price flexibility. The chance of bankruptcy dec reases as firms produce less, making this an attractive option during times of contraction. The risk of bankruptcy imposes a marginal cost on the firm that is taken into account in the production process, reducing output at each price level to cope. This c an be considered an agency cost which has the potential to negatively affect investment (Bernanke and Gertler 1989). This has the ability to drive downturns of the economy further. Any perception of an increase in risk leads aggregate supply to shift inwa rds. Thus, shocks to demand that result in recession likely lead to a decrease in aggregate supply which could affect the price level, although it could remain the same. Ultimately
19 in this view, price flexibility creates a sense of uncertainty that can amp lify bad economic situations and make them worse as demand and supply are interdependent. The emphasis on perception is very true to the Old Keynesian perspective and seems to hold a great deal of weight in the current economic downturn. In addition, New Keynesians have begun to explore some of the institutional behaviors that can affect the economy, particularly in regards to the credit market. During recessions imperfections in the credit market lead risk averse lenders to respond by shifting to safer a ctivities. In the midst of a negative economic shock, this can magnify bad conditions by raising the cost of financial intermediation, particularly through higher interest rates, often leading to credit rationing. This can put the economy in a downward spi ral, as borrowers constrained in the equity market find credit expensive and hard to attain, often resulting in firm bankruptcy. This part of the literature emphasizes the need to look beyond the traditional notions of the interest rate and the exchange ra te when looking at the business cycle ( Snowdon and Vane 2008 ). Looking at financial intermediaries, like banks, is extremely important as asymmetric information is problematic. Financial intermediaries, especially during recessions, are unable to assess cr edit worthiness due to imperfect information, and this helps drive economic downturns. Banks are risk averse firms that must use resources to screen and monitor to minimize the costs of uncertainty. This imposes a higher agency cost during times of recessi on, especially as overall net worth decreases, making asymmetric information problems of adverse selection and moral hazard more rampant in the credit market (Bernanke and Gertler 1989). The perception of uncertainty during
20 recessions is very high, and cre dit facilities can break down, hindering the growth of the economy. These problems of asymmetric information have definitely negatively affected the credit market in the recent recession. The suggestion that bank behavior can have an influence on broader economic fluctuations has been suggested, but not fully embraced. This is especially true when it comes to the liquidity preference of banks.
21 Chapter 2: Significance of Bank Liquidity Preference I. Why does bank behavior matter? The New Keynesian literature on the business cycle implies that frictions in financial markets can play a part in driving fluctuations in the economy. This suggests that there is a role for banks in the New Keynesian approach, although it is grounded in th e assumption that the banking industry is monopolistically competitive given imperfect information. This assumption hints that financial intermediaries are passive in their behavior, as they essentially take prices from their competitors. Ultimately the as sumption of monopolistic competition maintains the idea that banks are not able to significantly affect the economy through their behavior. While this may suggest that banks play a role in the business cycle, no attempt to explicitly implicate the behavior of banks with changes in the macroeconomy, especially in terms of liquidity preference, is mentioned in the literature. This thesis attempts to demonstrate that bank behavior has an It se ems as if the nature of the banking sector is more oligopolistic than monopolistically competitive, especially since the financial crisis. This has become much consolidation of banks. The banking industry has become controlled by a few firms, allowing each of them to have a much greater share of the market and sense of power in the industry. Banks are profit seeking and have become increasingly able to impose new rules on con sumers. Given the relative power of the small number of firms left in the banking industry, consumers have had little ability to push back on the new rules imposed on them as they have few alternative options regarding where they can put their
22 money. This has given banks a greater ability to do what makes them the most money, which has the potential to affect individuals and firms seeking loans considerably. As today most deposits are controlled by a handful of financial firms, these banks have control ove r a very significant amount of money in the economy. Banks then have the ability to decide whether this great amount of money will be lent out to circulate through the economy or held in reserves. Below certain levels of interest, banks seem to be holding increasing amounts of reserves, suggesting that banks decide when to lend and when not to given the rate of interest. As money in reserves has no real ability to help spur economic growth, bank behavior seems to be very significant, especially in recession ary periods. The decision by banks to lend or hold money in reserves is The decrease in the number of depository institutions in the financial sector has decreased competition in the industry. Subsequently profit driven banks are better able to do what they want to make them money without necessarily considering the overall lending for consumption and investment, the behavior o f banks in terms of their liquidity preference has become more important than ever to consider in terms of the business cycle. When banks hold money in reserves instead of lending or investing, their liquidity preference increases. This behavior is reflec ted by an increase in total reserves in the bank, which are their holdings of deposits in accounts at the Federal Reserve and cash holdings. Banks in the U.S. are required to keep a certain amount of reserves on transaction deposits. Reserve holdings above the required amount are considered excess
23 reserves. When gauging liquidity preference of banks it is necessary to look at total part that is most likely to change is excess liquidity given that banks can make the choice to hold whatever amount of excess reserves they want to. Excess liquidity in the banking sector has been a big topic in monetary macroeconomics lately, as banks are holding considerably more excess res erves than in the past, as is shown in Figure 2.1. Figure 2.1: Excess Reserve Holdings by Banks Excess liquidity is important in the discussion of liquidity preference because banks are not required by law to hold excess reserves. The increase in excess reserves to the extent shown in Figure 2.1 suggests that banks have been actively choosing to hold excess reserves instead of lending that money out as they typically did before 2008. and liquidity preference given a large increase in the money supply. The lost decade in
24 Japan is similar to the Great Recession in the U.S. in many ways as the burst of an asset bubble led to a crisis in the banking sector. In Japan this led to a ten year period of little to no economic growth. Despite growth in the Japanese money supply and extremely low interest rates, there was an emphasis in Japanese firms to pay down debt and save. This meant there was less spending in the economy. As banks held more money during the Lost Decade given low demand for funds, ideas like excess liquidity and liquidity preference have gained prominence in the literature (Koo 2009). However, the connection between liquidity preference, excess liquidity in the banking sector, and the broader economy have only been linked to inflationary expectations, especially in the New Keynesian literature. In both Japan and the U.S., monetary measures have been used to ease the severity of credit rationing by increasing the money supply greatly in an attempt to lower interest rates. The Bank of Japan has argued against the u se of monetary tools like quantitative easing, which has been used recently in the U.S. to decrease long term rates and pumped money into the economy. As these tools increase the money supply greatly, more money is likely to be held by banks in reserves as the economic outlook tends to be relatively bleak in times of recession. The major criticism from the New Keynesian perspective of this policy was that the inevitable increase in reserves in banks would stimulate inflationary expectations and consumer pri ces would increase, having little effect on the real variables of the economy (Krugman 2000). Other research holds that as the money supply increases through open market operations and the purchase of other long term assets, there are links to real yields and asset prices ( Eggertsson and Ostry 2005). T he main issue presented in the literature with regard to significant increases in
25 the money supply is the excess liquidity that accompanies it, only because it can supposedly spark high inflation as a recovery begins. Instead of assessing the extent to which the increase in bank liquidity would affect real variables in the economy, the literature only looked at what would happen to inflationary expectations as a result of the money growth. Thus, the importance of banks holding more money was not emphasized in the Lost Decade literature, as monetary seen as passive in their behavior and not active in driving changes in the b roader more money is being held by banks, it is unlikely that this will lead to an increase in consumer prices given that the money does not flow to producers or cons umers. Ultimately viewing fluctuations in the economy through the liquidity preference approach assesses to what extent money affects prices and the type of price. It does not assume that all prices automatically increase with growth in the money supply as is seemingly the big concern in prior literature surrounding this issue. II. The Influence of Financial Liberalization The relationship between bank behavior and the broader economy has become incredibly important in the age of financial liberalization. Financial liberalization emphasizes loosening government controls on the economy. The rise in prominence of this ideology occurred around the same time as the rise Real Business Cycle theory and the neoclassical revolution. Beginning around 1980, the poli cies associated with the ideals of financial liberalization gave banks a much greater sense of freedom, removing many regulations that were imposed during the Great Depression in the 1930s. As
26 financial liberalization sought to deregulate interest rates, a sset buying, and other facets of the banking industry, there have been significantly more financial crises in the past thirty years (Hellman n et al 2000). Financial crises often precede periods of economic contraction, as has been evident from the financi al crisis of 2008 onwards. Ultimately as banks have a greater ability to take more risk as they seek higher potential returns, the financial sector is more likely to face financial crises. As banks affect all kinds of consumption and investment, which are important drivers of the economy, the consequences of their actions tend to affect the entire economy instead of just one industry. Deregulation of the financial industry has made bank behavior that much more important to study when assessing the business cycle. III. The Backdrop of Recession The downturn of the sub prime housing market in 2007 marked the beginning of the Great Recession, a trough that we have not yet recovered from in the United States. The policy response has been unprecedented, as the Federal Reserve has significantly increa sed the money supply and a fiscal stimulus provided over eight hundred billion dollars to the ailing economy. Despite these policy measures, the recession has been difficult to emerge from as unemployment remains relatively high, and investment and lending remain low. One possible reason for this is that banks are holding a large amount of reserves instead of putting this money back into the economy to spur consumption and more importantly, investment. A brief look at Figure 2.2 shows a strange relationship between loans at commercial banks and excess reserves after 2007.
27 Figure 2.2: Excess Reserves in Commercial Banks and Total Loans While loans continued to increase in the beginning of the crisis as they had for the two decades before, lending took a sharp dip in the middle of the recession. Around the same time, excess reserves began to increase at an unprecedented rate. The lack of an immediate effect of the recession on both excess reserves and total loans indicates a lag in the sense that the rece ssion was not fully realized as soon as it began. It took time for economic agents including individuals, banks, and firms to recognize the downward turn of the economy. When they did, excess reserves continued to increase, and the total amount of loans in the economy decreased significantly from about 2008 into 2010. While lending has fluctuated since that initial downturn, it is clear that lending is still not where it was before the recession began, and that is significant. Clearly both the total amount of loans and excess reserves have fluctuated since the financial crisis, and this erratic behavior indicates a lack of confidence and certainty in the economy.
28 As the financial crisis emerged, excess reserve holdings by banks increased from the minimum am ount set by the required reserve ratio to almost 1.1 trillion dollars in a very short amount of time in 2008. At the same time, lending came to a halt (Gavin 2009). While there have been fluctuations in excess reserve holdings and lending practices on beha lf of banks since 2008, ultimately the initial shock started a trend that has persisted far after the financial crisis began as loans by commercial banks remain low and excess reserve holdings by these same banks remains high. Economic theory holds that an increase in reserve holdings should be offset with an increase in lending to combat the loss in income for the banks. However, credit rationing clearly occurred during the financial crisis. It is important to note that a downturn in credit can reflect a w eakness in demand for loans from a credit crunch that is supply induced. This means that demand for loans from individuals and firms might be weak because of a decrease in credit offerings that were initiated by the banking sector at the beginning of the f inancial crisis. The supply induced negative credit shocks could have lowered the perception of the state of the economy, which was driven down further on the demand side. Regardless credit rationing clearly occurred in the midst of negative perceptions, a s Figure 2.3 demonstrates.
29 Figure 2.3: Excess Reserves and the Money Multiplier Credit rationing becomes fairly clear given that the money multiplier decreased significantly as excess reserves increased in 2008, demonstrating that money was not flowing in the economy and was presumably resting in reserves instead. This idea is further supported by the dramatic increase in excess reserves. While the decline of the money multiplier could be brought on by a lack of demand for loans, the argument that this was brought on by a supply side shock in the banking sector seems to have some weigh t as is evidenced by the effect on the money multiplier. As the Fed increases the money supply, excess reserve holdings have the potential to increase significantly. When these reserves are not converted into loans, the money multiplier tends to lose its power. Thus bank liquidity preference, or the choice by a bank to hold money in reserves instead of lending it, has the ability to slow down the flow of money in the economy, ultimately restricting growth. In this light, bank liquidity preference should th eoretically have the ability to drive the business cycle.
30 IV. Why are banks holding so many reserves? Banks hold reserves both intentionally and unintentionally. It is important to examine why the recession has led to banks holding more in reserves and w hy this is an issue for the macroeconomy. If banks choose to hold money in reserves, they are failing to put cash back into the economy. If this behavior is not temporary and instead lasts for a while, the power of the central bank to help the economy alon g through growth in the money supply is undermined. The strength of monetary policy is extremely weakened when banks hold reserves on purpose for long periods of time. However, if the build up of reserves is due to outside factors that cannot be contro lled by the banks, there is nothing that can be done. The most even if funds are available. If both individuals and firms in the economy believe that the economic ou tlook is not favorable, they may be less likely to borrow, as they would be taking on risk without expecting an almost certain higher return. Consumers and businesspeople might decide to pay down debts, decrease inventories, or postpone investment during r ecessions as they would prefer to take risk on investments in a time that looked more economically advantageous. Thus, in light of extremely low interest rates, banks could potentially do nothing to decrease the amount of reserves and get the cash back int o the economy. A lack of loan demand is likely to be part of the reason why banks are holding so many reserves today. However, as the economic outlook in the U.S. continues to get better in the midst of better jobs numbers, demand for loans should pick up, making this reason alone insufficient in explaining why banks are holding so much liquidity.
31 Additionally, if the cost of converting bank cash into loans for consumers is too high, then banks may be sitting on reserves for good reason. The cost of chan ging reserves into income producing assets for banks can be very high. If the costs would outweigh the benefits, then profit maximizing banks would have to keep these cash assets in reserves rather than take the risk of losing money by lending. This tends to be true risk of lending the face value for a loan much higher. While it makes sense that a bank would not want to lend if it would cost more for them than it wo uld make, there is still choice associated with making loans or not. Banks calculate the costs associated with lending at a low interest rate by considering the cost of internal maintenance of converting the reserves into a loan, the cost of credit informa tion of the borrower, the interest rate (Frost 1971, Lindley et al 2001, Khemraj 2010). While these factors can help determine what the relative monetary cost of making the loan would be, the nature of factors like the cautionary risk of making a new loan make the choice to lend or not more subjective. Banks could be increasing their liquidity preference in order to decrease exposure to credit risk, allowing them to mit igate the costs associated with making risky loans to earn the interest payment. reserves is critically related to the uncertainty of flows in and out of its reserve accounts. It follows that in times of great uncertainty, like recessions, banks would hold more in reserves to account for the fact that changes in the required reserves are happening more
32 often, given run on banks by individuals and firms. The extent to which this happens is in part based on the size of the bank (Judson and Klee 2009). Uncertainty about risk exposure demonstrates the problems associated with asymmetric information. During recessions, asymmetric information problems tend to have large consequences. As the value of all assets fall during recessions, both consumers and firms have a tendency to be overleveraged. This makes it difficult for banks to identify quality borrowers, making these institutions more wary of lending (Ball 2008). Additionally, as took on weak aspects of other institutions. As they deal with the internal ramifications of those mergers and must find a way to decrease their leverage, they may be less likely to focus on lending. As a result of weaknesses surrounding the financial crisis, banks often try to decrease the size of their balance sheet. Typically banks will call in loans for repayment to create liquidity, meaning that loans that could have rolled over into the next period are called in abruptly upon their maturity. They will also sell loans to other banks to decrease the asset side of the balance sheet. In order to balance that action on the balance sheet, they would subsequently decrease their liabili ties by reducing deposits. Reducing liabilities would have to be done through a decrease in excess reserves given that banks must hold a required amount of reserves (Keister and McAndrews 2009). This kind of behavior is targeted by many of the Federal Res erve lending programs, in which the Fed has tried to stop banks from calling in or selling loans because this behavior would definitely lead to a contraction in business investment in the economy. In order to combat this behavior, the Fed provides a credit
33 account to stop the individual bank from reducing its loans. The distress caused by the uncertainty around recessions makes banks more likely to hold this credit in reserves so that they have enough cash on hand to deal with potential losses in the future. Once the money is pumped into the individual bank, that institution has the ability to decide whether or not those funds will stay within the bank or not. Some ideas about banks voluntarily holding reserves emerged from the Great Dep ression. One idea is fairly prevalent in the older literature that holds that banks might hold more total reserves to prove their solvency during difficult economic times. They may need to demonstrate their ability to pass a stress test and weather the nex t potential storm (Morrison 1966). However, the extent to which reserves are held is never established in this literature. The amount of reserves being held reflects a choice by the banks. Finally, banks intentionally hold reserves when they have sound c apitalization and can use the excess reserves, including money borrowed from the Fed, to purchase assets that have a higher or more stable return than lending. There could be a speculative aspect to holding the money, in which banks would prefer to purchas e relatively cheap assets in a recession instead of lending out a full amount to a borrower. This boosts the asset side of the balance sheet, and the assets can likely be sold for a higher price in the future, making the bank more profit. Along these line s, in October 2008 the Fed took steps to start paying banks for holding excess reserves. The rationale for this action is that it gives the central bank the opportunity to maintain influence on interest rates independent of the excess liquidity created. Th en the size of the money supply is determined by the state of the financial
34 sector, and the central bank can continue to set interest rate targets based on what is happening in the macroeconomy (Keister and McAndrews 2009). While the ability to separate th e growth of the money supply from the interest rate was important for monetary policy during the recession, the fact that interest, even as a small amount, continues to be paid on excess reserves over three years later seems to have incentivized banks to h old excess liquidity even more. Some Fed policies have seemingly encouraged banks to sit on money instead of lending it, which is signified by the large amount of reserves being held today. The reasons why banks are exhibiting an increase in liquidity pre ference is due to the effects and responses to the financial crisis and subsequent recession. While banks are seemingly holding reserves both intentionally and unintentionally, all of these reasons have led to a significant increase in total reserve holdin gs by banks. This has resulted in a positive shock in bank liquidity preference, which is an institutional behavior that has likely had a significant impact on the economy given that it has caused less money to flow throughout the economy. V. The Importan ce of Bank Liquidity Preference This thesis adds to the literature because it looks at the importance of institutional behavior in the economy in a new light as the focus is on shocks in liquidity in the banking system and how these changes in reserves aff ect the broader economy. Monetary policy has attempted to increase the money supply by unprecedented amounts in the last few years through the initial monetary stimulus and through subsequent programs like quantitative easing. These programs were meant to drive down the interest rate to increase consumption and lending, as traditional Keynesian theory suggests. However,
35 liquidity in the banking sector has grown very significantly, implying that instead of lending, banks are sitting on reserves. This thesis explores some of the reasons behind an increase in bank liquidity preference. There are generally three main schools of thought The first is from Morrison (1966) and holds that banks increased reserve holdings slowly during the 1930s not because of low interest rates but because of shakiness in the banking sector and subsequent government regulation which forcibly increased reserves. Given the parallels between the Great Depression and the Great Recession, this argument can plausibly be made today as wel l. The second, from Friedman and Schwartz (1963), maintains that when the money supply increases there is essentially an automatic increase in liquidity holdings as changes in the stock of money move the economy. Finally, Frost (1971) supposes that banks i ncrease liquidity preference to decrease the costs associated with reserve requirements, as the banki ng sector is fraught with random changes in reserve flows and transactions costs. As such, there is a kinked demand curve at low rates of interest, as bank s begin to hold increasing amounts of reserves. While the Frost paper is concerned with treasury bills, this same idea seems most applicable to this thesis in terms of why banks are holding an un precedented amount of reserves because clearly the interest r ate plays a role. This becomes evident in the model for bank liquidity preference in the methodology chapter. Regardless of the reasons behind banks holding such liquidity, the changes in bank behavior in recent years through an increase in bank liquidity preference has likely hurt the economy as very liquid money is available but is not being spent to stimulate
36 economic growth. As banks have control over this money to at least some extent, their liquidity preference is likely to impact the economy signifi cantly. economy contracts to an even greater extent, driving the business cycle furth er down. This means that as bank liquidity preference increases or decreases, key economic indicators will be affected, which demonstrates an effect on the aggregate level. This thesis will assess the relationship between changes in bank liquidity preferen ce and macroeconomic variables including GDP, unemployment, inflation, the stock market, housing starts, and corporate profits after tax. Ultimately, changes in bank liquidity preference should help drive economic fluctuations.
37 Chapter 3: Methodology I. Model of Bank Demand for Reserves In order to gauge the extent to which shocks in bank liquidity preference affect the overall state of economy, there must be a way to assess when and the extent to which banks change their liquidity preference. In order to do that, there are a few steps to be taken methodologically. First, data pertaining to total reserve holdings, the loan rate, and the federal funds rate were gathered from Federal Reserve Economic Data of the Economic Research depar tment of the Federal Reserve Bank of St. Louis, commonly known as FRED. Quarterly data on total reserves in depository institutions, the bank prime loan rate, the federal funds rate (FFR) were used from the first quarter of 1980 to the fourth quarter of 20 11. 1980 was chosen given that it signifies the beginning of financial liberalization, which marks a significant change in bank behavior. The fourth quarter of 2011 was chosen as the endpoint because it is the latest data available given all of the variabl es being considered in the model. The loan rate was chosen because it is determined by banks themselves and is thus reflective of their liquidity preference. The FFR was also chosen because it is the interest rate between banks when they lend and borrow t o each other. The FFR sets the foundation for all interest rates in the economy and is an important benchmark in the economy as banks commonly borrow and lend to and from each other. The data from FRED indicates that while the FFR was near zero at approxim ately 0.07, the loan rate was well above the FFR at a steady 3.25 by the end of 2011. This difference and
38 fluctuations in both of the rate makes it important to judge total reserves against both the FFR and the loan rate. The relationship between the loan rate and total reserves is central to study today given that the credit market is seemingly in a liquidity trap like situation. No matter how much the money supply incre ases, the loan rate does not decrease. Thus, the money supply and monetary policy do not stimulate economic growth once the loan rate hits its minimum level. The loan rate does not decrease because banks can effectively determine what this loan rate will b e given the costs associated with funds, risk, and liquidity conditions. At this minimum threshold, the loan rate becomes equal to the marginal cost of holding cash and the marginal cost of making loans. At lower loan rates, the marginal costs exceed the p otential benefits of lending, so banks are more likely to hold more reserves, increasing their liquidity preference. T he opportunity cost of holding reserves becomes lower as the loan rate decreases, especially in the midst of asymmetric information proble ms that exist during recessions (Khemraj and Proao 2011). The relationship between total reserve holdings at depository institutions and the effective FFR is important for similar reasons. As the loan rate is determined by banks, the federal funds rate a cts much in the same way, as banks are the borrowers and lenders and can negotiate what these rates will be. Banks will only lend at a rate that they feel is worth not holding the money in reserves instead. Essentially the relationship between excess reser ves and both the loan rate and the effective FFR are as follows. As either of these interest rates decrease, reserve holdings increase, meaning that liquidity preference increases as is explained above. As either of
39 these rates increase, reserve holdings d ecrease, meaning that liquidity preference decreases. This is because the opportunity cost of holding cash is too high and more can be made by lending out reserves at that higher rate. This indicates that the shape of the bank liquidity preference is downw ard sloping. It becomes asymptotic at the minimum threshold interest rate given that below a certain rate, banks start holding more reserves instead of lending. At the rate threshold, the bank liquidity preference demand curve becomes essentially horizonta l because changes in the rates downward encourage banks to hold as many reserves as they want to. Accordingly, the curve would look as follows with on the x axis standing for total reserves, on the y axis indicating the interest rate (either loan rat e or effective FFR), representing the minimum threshold rate, and indicating the liquidity preference curve for banks. Figure 3.1: Model of Bank Demand for Total Reserves
40 Based on this information, the banks demand for excess reserves, or bank liquidity preference, can be modeled as: (1) where, = rate of interest (loan rate or FFR) = minimum threshold loan rate = coefficient for total reserves = total reserves = residual. Total reserves are expressed as a reciprocal to reflect the asymptotic nature of the bank liquidity preference curve. This model provides an equilibrium liquidity preference for banks as is demonstrated by the R The most important part of the model, however, is the residual of the model, noted as As the model considers the interest rate as a function of excess reserves, the residual reflects the difference between the observed value in a given quarter and it s equilibrium value. Thus, the residual demonstrates liquidity preference shocks for banks in every quarter being considered. As the residual goes above zero, bank liquidity preference increases. As the residual falls below zero, bank liquidity preference decreases. This is demonstrated in Figure 3.2 for the relationship between total reserves and both the loan rate and the FFR. Figure 3.2 is the residual output of the model of bank demand for reserves and serves as a useful tool for seeing how banks change their liquidity preference.
41 Figure 3.2: Residual Shocks Using Loan Rate and Federal Funds Rate The changes in reserves for banks clearly differ when considering the loan rate and the FFR separately. While the FFR follows a fairly steady path most of the years, the loan rate regression shows much more fluctuation. This indicates that bank liquidity preference is very responsive to changes in the loan rate and less responsive when it comes to changes in the FFR. Around 2008, bank liquidity preference starts to become more responsive to both rates. There tends to be a decrease in the residuals, or a decrease in bank liquidity preference, during recessions. For the loan rate t his behavior is evident during the recession in the early 1990s, the early 20 00s, and in 2008 as well. For the FFR, this behavior is particularly evident after 2008. This makes sense given that banks tend to lose liquidity during recessions as expectations of the economy weaken, forcing banks to shift reserves to cover issues on th eir balance sheets. Additionally, people tend to withdraw more money from banks during these times, which would lead banks to have less in reserves. Liquidity preference tends to increase after these times, as expectations
42 remain low, especially for the lo an rate. This seems to be less of the case with the FFR since 2010, which is an interesting result. Generally b anks want to be able to have the reserves necessary to deal with another possible downturn. Furthermore, banks may be looking for stronger invest ment opportunities after recession. The following scatterplots demonstrate that this model is a good representation of what is actually happening quarterly from 1980 to 2011 between both loan and federal funds rate and total reserves as seen on the left, as well as the federal funds rate and total reserves, as seen on the right. Figure 3.3: Empirical Evidence of Bank Demand for Total Reserves The intuition behind the asymptotic nature of the curve seems to be supported by its actual shape as is demonstrated with a line of best fit in Figure 3.3. When considering the relationship between both types of interest rates and the total amount of reser ves held by banks, the liquidity preference curve for banks becomes level at a particular interest rate. The lower bound loan rate seems to hover just below 4%, while the federal funds rate seems to be asymptotic just above 0%. The loan rate threshold is i ntuitively higher than the federal funds rate as banks are profit seeking. It seems probable that the loan rate
43 has a greater effect on variables that are directly affected by banks, like mortgage accessibility and thus housing variables. The federal funds rate is likely to have a greater effect on broader macroeconomic aggregates like GDP Bank liquidity preference shocks are central to this thesis. When the residual is positive for a given time period, this indicates that bank liquidity preference has in creased. When this happens, the bank liquidity preference curve shifts upwards. In the case of the negative residual in a given quarter, liquidity preference has decreased, indicating a downward shift of the bank liquidity preference curve. Drawing from Fi gure 3.2, when the residual goes above zero the curve shifts rightwards. When the residual goes below zero, this would indicate a leftward shift of the curve. This can be demonstrated by the stylized model in Figure 3.4. Figure 3.4: Model of Bank Liquidity Preference Shocks
44 Figure 3.4 indicates that a leftward shift from to is a negative liquidity preference shock, meaning that banks have a lower liquidity preference, and there are less reserve holdings in the depository institutions. The rightward shift from to indicates a positive liquidity preference shock for banks. This means that banks have a higher liquidity preference, and they thus hold more in total reserves. This thesis seeks to identify when these shocks happen and to what extent they affect macroeconomic fluctuations. II. Model of the Effect of Bank Liqu idity Preference Shocks on Macroeconomic Indicators Once the residual information is attained for every quarter using the program EViews, changes in liquidity preference, positively and negatively, can be assessed. The residual information subsequently b ecomes the data on bank liquidity preference shocks. In order to see whether or not these shocks have an effect on economic trends in the macroeconomy, bivariate regressions are run to examine the relationship between individual key economic indicators and bank liquidity preference shocks. In order to do this, an ARDL model is used. The approach of using the total reserves demand model and then ARDL is adopted from other papers that assess the sources and effects of excess liquidity across the world (Fieldi ng and Shorthand 2005, Saxegaard 2006, Khemraj 2010). While this thesis is considering the importance of shocks in bank liquidity preference, the same types of models can be used given the similarity of variables and need for lags in explaining fluctuation s of macroeconomic variables. The ARDL model is an error correction model that finds a stable long term equilibrium between two variables. In order to use the ARDL model, the data must be
45 stationary, meaning that there is no unit root or linear trend. Sta tionary data is important because many of the assumptions of time series statistical forecasting are premised on the constant over time. If the data is not stationary then the difference of the variable must be used, which can be achieved by calculating the log of each data point of the macroeconomic variable. The AR stands for autoregressive, which are models that include lagged dependent variables. The DL stands for distributed lag, which includes lagged independent variables. It is important to use lags because time will elapse from the time the liquidity shocks occur to when the economy acknowledges that banks are holding more or less money, or more or less money is circulating through the economy (Asteriou and Hall 2007). This dynamic model can capture the fact that sometimes a current time period. In terms of the number of lags ne cessary, the data helps guide how many lags are relevant in the model, although there should generally be no more than three. The general ARDL model looks as follows, beginning with the following equation, (2) where the first half of the equation reflects AR model and the second the DL model. A residual is included in the model, although as the overall ARDL approaches a single term the residual, or random te rm, drops out. Thus the ARDL model approaches
46 Substituting these equilibrium values back in the general equation, the following equation is formed: (3) Solving to put and on either side of the equation, the next step looks as follows: (4) Finally solving for in terms of the coefficients, the following long term coefficient is approached: (5) The long term coefficient solved using this equation becomes the following: (6) The long term coefficient provides an equilibrium value that allows a relationship between the independent and dependent variable to be assessed. As is on the left side of (5) and is on the right side, the long term coefficient gives a sense of how the independent variable affects the dependent variable (Baddeley and Barrowclough 2009). The independent variable in all of these bivariate regressions is bank liquidity preference shocks, while the dependent variables are macroeconomic indicators like G DP. is the impact coefficient, which is the weight attached to the current independent variable in the current time period. is especially important given that it shows how much the average change in the dependent variable ( ) will be when the in dependent variable ( ) changes by one unit. Both the shock variables and the dependent variables must be lagged to assess if there are significant relationships between liquidity preference shocks in previous and current time periods and dependent var iables in previous and current time periods. It is
47 important to evaluate which coefficients are seemingly significant and which are not using the EViews software. To ensure that there is no autocorrelation or heteroskedasticity that would undermine the sig nificance of the results, the Newey West HAC estimator is used to correct for this and present non biased standard errors. This is important in time series data given that the residual tends to be correlated over time due to non standard variance of the da ta, which tends to misconstrue standard errors. A Wald F test is used to restrict the model by taking all of the seemingly insignificant coefficients and setting them equal to zero, such that, for instance, Thus in this example the null hypothesis holds such that H 0 : The alternative hypothesis holds that at least one of these coefficients does not equal zero, such that H 1 : at least one coefficient among H 0 can only be rejected when the F statistic is greater than the F critical, or the p value is less than 0.05. As long as the null hypothesis is not rejected, the insignificant coefficients can be thrown out of the final model and the significant ones can rem ain because the insignificant coefficients would all feasibly equal zero. From there, the long term coefficient can be calculated using the significant coefficients, which provides a numerical relationship between liquidity preference shocks and the depend ent variable in the given bivariate regression. The impact coefficient is especially important in gauging the relationship between the liquidity preference shocks and the changes in the macroeconomic variables as well. III. What is the likely effect of liquidi ty preference shocks on key macroeconomic indicators? The model presented in the thesis seeks to assess the impact of bank liquidity preference shocks on macroeconomic variables including GDP, inflation for both
4 8 consumers and producers, unemployment, corp orate profits after taxes, and housing starts. These variables were chosen because they are important in getting a sense of the strength of the economy and its fluctuations. The bank liquidity shock model is at the core of the thesis because changes in ban k behavior in terms of their liquidity is hypothesized to have an effect on most of these variables, suggesting that shocks in bank liquidity preference help drive fluctuations in the economy. It is important to establish why these individual variables are chosen and the likely effect of liquidity preference on them. i. Unemployment Unemployment is an important variable to look at when gauging fluctuations in the economy because of its relationship with output and its significance in driving expectations in the economy both positively and negatively. Unemployment is related to output because generally when more is being produced, more people are employed. When more people are employed and the economy is operating closer to its natural rate of unemployment, expectations about the economy are generally good. This has been evident recentl y as jobs numbers have improved, especially in the private sector, allowing the economy to slowly but pretty surely emerge from its depressed state. Ultimately as banks hold money in reserves, and the money is not flowing through the economy via consumptio n or investment, less people are likely to be employed. This suggests a positive relationship between a positive bank liquidity preference shock in which banks hold more in reserves and an increase in unemployment. Conversely, as banks let go of some of th is liquidity, especially through lending, unemployment is likely to decrease.
49 The effect of liquidity preference shocks on unemployment is likely to be more powerful when considering the loan rate. This is because the loan market seems incredibly importan t in getting people to consume more and businesses to invest more given that this is somewhat dependent on greater lending. However, given the fact that unemployment is a major macroeconomic aggregate, it is likely that the FFR would have a strong impact o n unemployment as well. ii. GDP Gross domestic product (GDP) is chosen because fluctuations in the economy and the business cycle in general are typically measured by changes in output. The relationship between shocks in liquidity preference and GDP seems som ewhat unclear because of the fact that the government can bolster the GDP by spending when the economy is relatively weak (Krugman and Wells 2007). Intuitively it would make sense that an increase in bank liquidity preference would mean that banks are holding more money in reserves and thus there is less money to flow in the economy. If consumption and investment are lacking for these reasons, the government is likely to step in and increase spending as it did in 2008 to help the economy weather the storm. While an increase in liquidity preference should lead to a decrease in GDP, it may not. If there is an effect of liquidity preference shocks on GDP, it is likely to be more powerful in terms of the FFR given that it is a broad based macroeconomic variable. iii. Inflation The consumer price index (CPI) is chosen to gauge changes in prices for consumers given changes in bank liquidity preference. Th is is a key indicator in the economy because changes in consumer prices have serious potential to drive what is
50 power. The producer price index (PPI) is also chosen to see t he effect of this behavior on producer prices, which are also important because the cost of inputs on the supply side have the ability to increase or decrease supply, which can have major ramifications in the economy too. Economic theory indicates that an increase in liquidity in the banking sector would lead to inflationary expectations. The theory holds that when banks are holding more liquidity, this money should eventually make its way into the economy, which ultimately results in a rise in general pric es especially if it happens in a short period of time. A rise in prices would indicate a positive relationship between bank liquidity preference and the CPI and PPI. One of the major criticisms of increasing the money supply by the Fed was that hyperinflat ion could occur by pumping such a huge amount of money into the economy. However, the fact that banks are holding a great deal of money today in reserves suggests that the money has not yet entered the economy and indicates that this behavior has not affe cted consumer or producer prices. This means that there should not be a significant relationship between bank liquidity preference shocks and both CPI and PPI. This should be seen both in terms of the loan rate and the FFR. iv. Corporate Profits after Taxes C orporate profits after taxes are an important macroeconomic indicator as well. When business is booming, it indicates that the economy in general is doing well. When banks are holding more liquidity it means that they are holding money instead of lending i t. This behavior suggests that corporate profits will likely decrease as banks increase their liquidity preference. If corporations are not able to access credit that could enable
51 them to increase their output capacity, they are not likely to make as many profits. This suggests that there is a negative relationship between bank liquidity preference shocks and corporate profits after taxes. The effect of bank liquidity preference shocks on corporate profits should be greater in terms of the loan rate versus the FFR given that the opportunities that can help generate growth in the economy. A decrease in corporate profits may seem somewhat hard to advocate for today giv en that like banks, many corporations have seen record profits and major cash inflows despite the depressed state of the economy. The discrepancy between the hypothesis in this thesis and real life may be because of a few factors. First of all, the biggest and most well known corporations are likely to have had little issue accessing credit given their proven historical record regardless of bank liquidity preference. It is smaller corporations and start ups that are more likely to hurt because of the lack o f lending on the part of banks. Additionally, many of the big corporations are seeing positive shocks in their own liquidity preferences because instead of investing, they too are holding onto more money than ever. v. Stock Market The stock market is anothe r important macroeconomic variable that has some ability to gauge fluctuations in the economy. The state of the stock market is assessed from the vantage point of both the Dow and the S&P. The Dow Jones Industrial Average is a stock market index that is he lpful to look at given that it is made up of thirty WalMart, among others (Dow Jones Averages). When these major American companies
52 are doing well, it generally means good things for the economy, meaning that this index is an important indicator of the state of things. The same can be said for the S&P 500 Index, which is similar to the Dow but includes many more companies. The stocks in the S&P include large publicly held co mpanies that trade on either the NASDAQ or the New York Stock Exchange (S&P 500). Even though the stock market has been incredibly volatile recently as evidenced by both of these indices, generally when the stock market is doing well, sentiments about the overall economy are more positive, helping it move in the right direction. As banks hold more money, it is likely that as consumption and investment are seemingly worse off, meaning that the stock market should not necessarily be doing as well. This sugges ts a negative relationship between a shock in bank liquidity preference and both the Dow and S&P. However, given the speculative nature of the stock market, there actually could be a positive relationship between bank liquidity preference shocks and the mo vement of the stock market. As banks hold more money, the idea is that this money must flow into the economy eventually, and it is likely to do so to many of the companies represented by the index. This may make investors more likely to invest in these com panies, improving both the Dow and the S&P. Given the importance of investment of major businesses in driving fluctuations in the economy, it is likely that the loan rate will have more of an effect on the impact of shocks in bank liquidity preference on b oth the Dow and the S&P. vi. Housing Starts Finally, housing starts are investigated in this model. The housing market was very influential in bringing down the economy is 2008. It continues to drag down the economy, as less people are moving into homes and ev en less are buying new ones.
53 Housing starts are constructs of new homes (Federal Reserve Economic Data). When people buy homes it is a good indicator for the economy as they seem willing to buy big ticket items at what they consider to be the lowest price. This means that the housing market has hit its bottom, indicating that the economy is getting better. As new homes are being built, this shows that economic growth is progressing. However, this remains to be seen, especially because accessibility to credi t for people looking for mortgages has been very hard to come by in recent years. Additionally, construction has almost come to a halt, especially in areas worst hit by the housing bubble meltdown. Thus, intuitively it makes sense that as banks are holding more money, this prevents companies from building new homes and consumers from being able to access the mortgages they need to move into new homes. As demand for new houses has decreased because of a seeming lack of credit accessibility, this has decrease d the amount of new homes being built. This indicates a negative relationship between bank liquidity preference shocks and housing starts. This relationship would seemingly be more robust when considering the loan rate given the importance of the mortgage market in moving the housing market along. In the next chapter, these hypotheses will be considered in light of results of the regressions.
54 Chapter 4: Results and Discussion I. Effect of Bank Liquidity Preference Shocks on Macroeconomic Variables Using Long term Coefficients The effect of bank liquidity preference shocks on the chosen macroeconomic variables can be measured by calculating the long term coefficient. The long ter m coefficient is essentially a long term equilibrium value that assesses the relationship between the dependent and independent variables, as is explained in the methodology chapter. The long term coefficient is used in many error correction models in the macroeconomics literature and results are generally summed up in tables (Akitoby et al 2004, Schettkat and Langkau 2008). The long term coefficient is calculated by dividing the coefficients of the lagged shock variables by one minus the coefficients of th e lagged macroeconomic variables. In terms of the model of bank liquidity preference shocks, the long term coefficient evaluates the effect of bank liquidity preference shocks, the independent variable, on the dependent variables like unemployment and GDP. The nature of the relationships between bank liquidity preference shocks and the numerous macroeconomic variables were considered in the methodology chapter, and hypotheses were made for each dependent variable. This chapter will seek to confirm or negat e the hypotheses made in the previous chapter and discuss why the intuition behind these hypotheses likely worked out or not. It is always important to consider both the statistical and economic significance of the results provided by a model. While statis tical significance is helpful in demonstrating what is worth considering as significant, economic intuition should be able to back up the statistical significance (Ziliak and McCloskey 2004).
55 Table 4.1 summarizes the long term coefficients found for each macroeconomic variable in terms of both the loan rate and the FFR liquidity shocks. The table includes the long term coefficients and also demonstrates the number of significant lagged independent and dependent variables. The dependent lags are the lags of the macroeconomic variables themselves. If there are significant dependent lags for unemployment for instance, this means that unemployment figures in past quarters have an effect on unemployment today given the shocks in liquidity preference. The more im portant thing to consider are the independent lags, which are the lags of the shocks in regards to either the loan rate or the FFR. These reveal that liquidity shocks in prior quarters or in the current one have an effect on the given macroeconomic variabl e. Table 4.1: Long Term Coefficients Macroeconomic variables Long term coefficient estimate using the loan rate Long term coefficient estimate using FFR Unemployment 0.589321 ; three significant dependent lags, two significant independent lags 0.56911 ; three significant dependent lags, one significant independent lag GDP N/A; no significant independent variables N/A; no significant independent variables CPI N/A; no significant independent variables N/A; no significant independent variables PPI N/A; no significant independent variables N/A; no significant independent variables Corporate profits after taxes 0.80587 ; one significant independent lag 0.699 ; one significant independent lag S&P 0.214119 ; one significant dependent lag, one significant independent lag N/A; no significant independent variables
56 Dow 0.203861 ; one significant dependent lag, significant impact coefficient, one significant independent lag 0.223572 ; one significant dependent l ag, significant impact coefficient, one significant independent lag Housing starts 0.26888 ; two significant dependent lags, one significant independent lag 0.52301 ; one significant dependent lag, one significant independent lag For each macroeconomic variable, an ARDL (3,3) regression was run in EViews such that there were three lags of the dependent variable, or the macroeconomic variable, and three lags of the independent variable, or the shocks. An ARDL (3,3) was run twice for each mac roeconomic indicator, first considering the liquidity preference shocks in terms of the loan rate, and then the shocks in terms of the FFR. Thus there were two initial regressions performed for each macroeconomic variable. From there, statistically and eco nomically significant variables were considered, and a Wald F test was used to restrict the models such that only significant variables were used in calculating the long term coefficient. In the case that there were no significant shock variables, no long term coefficient is found as there is seemingly no relationship between liquidity preference shocks and those macroeconomic variables given the type of interest rate. In order to discuss the results demonstrated in Table 4.1 and further results that becam e evident when running the regressions, each macroeconomic variable is considered separately. i. Unemployment From the previous chapter, it was hypothesized that there is a positive relationship between unemployment and shocks in liquidity preference with t he loan rate having a greater impact on the effect than the FFR. Both of these ideas are supported by the long
57 term coefficients found, as the loan rate long term coefficient is slightly higher than the FFR long term coefficient. Both are positive, suggest ing that an increase in bank liquidity preference does lead to an increase in unemployment. This is likely because as banks hold more in reserves, less money is flowing through the economy through consumption or investment, both of which help create jobs i n the economy. Conversely, as banks hold less money and their liquidity preference decreases, more money is entering the economy, allowing unemployment to decrease. Unlike most of the other variables, the unemployment data did not have a unit root, meanin g that the data was stationary and did not need to be logged to be considered. In terms of the loan rate regression, significant lags were found for the three unemployment lags as well as the first and second shock lags. The impact coefficient, or shocks i n liquidity preference in the current time period was not found to be significant. Using the Wald F test, the insignificant impact coefficient and third unemployment lag were thrown out of the model as the null hypothesis of these coefficients equaling zer o could not be rejected. The restriction of the model allowed the long term coefficient to be calculated using the significant coefficients. In terms of the FFR regression, significant lags were found for the three unemployment lags as well as the first u nemployment lag. The impact coefficient and second and third shock lags were removed from the model, as the Wald F test confirms, allowing the long term coefficient to be calculated using the significant coefficients. It is interesting to note that the R 2 term for both regressions is 0.982, which is quite high. This is considerably higher than the R 2 terms for the other regressions. This suggests that the independent variable has a good ability to drive changes in the
58 dependent variable, implying that the shocks in liquidity preference given both the loan rate and the FFR have a strong ability to change unemployment. High R 2 terms are not a norm in social science data given the behavioral aspects, so this was an interesting result. All of the significance f ound in these results demonstrates that bank liquidity preference is important in driving changes in unemployment, which has significant implications for the business cycle. ii. GDP The hypothesis presented in the methodology chapter for the effect of shocks in bank liquidity preference on GDP stated that while there could be a positive relationship between these variables, this might not be the case given the role of government spending in GDP. As there were no clear significant shock lags for either the loan rate or the FFR regression, the results suggests that there may be some truth to the hypothesis. When the initial ARDL (3,3) was run for the loan rate shocks, the first and second GDP lags were found to be significant, which intuitively makes sense. Howe ver, the shock variables were all found to be very insignificant, with p values ranging from 0.16 to 0.76. When the ARDL (3,3) was run for the FFR shocks, there seemed to be some possible significance in the second and third FFR shocks given that their p v alues were 0.08 and 0.09 respectively. The third GDP lag, the impact coefficient, and the first shock lag were attempted to be thrown out of the final model using the Wald test. However, the Wald F test did not allow the model to be restricted to the seemi ngly significant variables, given that the null hypothesis, which sets all of the insignificant coefficients equal to zero, was rejected with a p value less than 0.05. If they were truly insignificant the null hypothesis would not be rejected, as the insig nificant coefficients should go to
59 zero. It makes sense that there might be more significance in the shock variables in terms of the FFR than the loan rate given that GDP is an aggregate and its components are affected by many kinds of different interest r ates, all of which are based on the FFR in one way or another. Given the difficulty in restricting the model and the lack of clearly significant shock variables, no long term coefficient was calculated for either the loan rate or FFR regressions, suggesti ng that there is no significant relationship between shocks in liquidity preference and GDP. While shocks in liquidity preference are likely to affect consumption and investment, when there has been a decrease in these variables since 1980, the government has stepped in to increase spending to lessen the blow that these negative effects can have on the broader economy. It is possible that the effect of an increase in liquidity preference by banks on GDP can be buffered by an increase in government spending, as the downturn in GDP is likely not as severe. iii. Inflation Inflation is considered by both the CPI and PPI in the model. Both yield similar insignificant results, as was hypothesized in the methodology chapter. This suggests that bank behavior in terms of liquidity does not help drive inflation on either consumer or producer prices, meaning that bank liquidity preference does not seem to affect real prices in the economy. When banks are holding more liquidity, money is not flowing into the economy, mean ing that it has no effect on prices on either the demand or supply side. Conversely, as banks decrease their liquidity preference and hold less in reserves, this action also seems to have no affect on these measures of inflation because expected inflation should match actual inflation as the prices are being realized.
60 The CPI data demonstrated a unit root, indicating that the data was not stationary. The log of the each data point was taken to take care of this problem in the model. Logging the data allowe d the regression to be done looking for a relationship between growth in the CPI and shocks in bank liquidity preference. When the ARDL (3,3) was run with the loan rate shocks, there were no significant shock lags. All of the p values were well above 0.05, ranging from 0.25 to 0.99. The first and third CPI lags were significant, but the lack of significant shock coefficients meant there is seemingly no relationship between these variables and thus no long term coefficient. When the ARDL (3,3) was run with the FFR shocks, the second shock lag was significant, as was the first lag in the CPI. However, when the Wald F test was used to restrict the model to the seemingly significant variables, the null hypothesis of the insignificant variables being equal to ze ro was rejected, making it difficult to find which variables were significant or not. As the model could not be restricted, no long term coefficient was found. The FFR seems to yield more seemingly significant results, which makes sense given that inflatio n is a broad aggregate, and prices tend to be related to many different interest rates in the economy. The PPI regressions were very similar. The PPI data showed that there was a unit root present, so the data was logged making the variable more represent ative of growth in the PPI. For the ARDL (3,3) in terms of the loan rate shocks, there were no significant lags. P values ranged from 0.33 to 0.56. Thus even though the first and second PPI lags show significance, no long term coefficient was found. In ter ms of the shocks in FFR, there were also no significant shock lags, with p values ranging from 0.17 to 0.42. These
61 results demonstrate a lack of relationship between growth in the PPI and shocks in bank liquidity preference. iv. Corporate Profits after Taxes T he hypothesis from chapter 3 holds that as bank liquidity preference increases as indicated by a positive shock, corporate profits will decrease. This indicates a negative relationship between the variables, which is supported by the long term coefficient calculated. Additionally, the idea was that the loan rate would have a greater effect than the FFR in terms of this variable, as corporations are very dependent on loans for growth and thus more profit. This idea is also confirmed by the long term coeffici ents, as the loan rate coefficient is more negative than the FFR coefficient. The corporate profits data demonstrated a unit root, so the data was logged to make the data stationary, making the dependent variable the growth in corporate profits after taxe s. Significance was not found in any of the corporate profit lags, but there was significance found in both the second lag of the loan rate shocks and the second lag of the FFR shocks. This formed the numerator of the long term coefficient calculation, and given the lack of significant corporate profit lags, it was set over one. A Wald F test was performed which allowed the model to be restricted to this one significant variable for both the loan rate and FFR regressions. It is an interesting result that th e second lags of the shocks in each case were significant, but the other shocks were not, as this essentially means that liquidity preference shocks from two quarters previous are affecting corporate profits, but not the contemporaneous shocks, the shock o ne quarter before, or the shock three quarters before. This suggests that changes in bank liquidity preference take some time to make reverberations in the economy, which makes sense given that money flows
62 through multiple mechanisms. This is why lagging t he variables in important, especially when it comes to changes in the movement of money in the economy. v. Stock market Both the S&P and the Dow were used as proxies to assess how shocks in bank liquidity preference affect the stock market, which is a broad macroeconomic variable that shows important fluctuations in the economy. It was hypothesized that an increase in bank liquidity preference could lead to either an increase or a decrease in the growth of the stock market. Clearly all of the long term coeffi cients for both the S&P and the Dow show positive numbers, suggesting that an increase in bank liquidity preference leads to an increase in growth in the stock market. When there is a positive shock in bank liquidity preference, there are expectations as t o what will happen with the money being held today. As investors believe that this money will eventually leave the banks through loans to companies, they may be more willing to invest in quality stocks today. On the flip side when there is a negative shock in bank reserves, this may indicate that banks are in distress or are overleveraged, which could hurt expectations for the market, driving its growth down. There is a speculative nature to banks holding more or less money that seems to have an impact on m ore speculative macroeconomic indicators, like the S&P and the Dow. The idea that the loan rate would be more robust holds true for the S&P but not the Dow. This is a somewhat interesting conclusion given that the S&P is a much broader index than the Dow, which would seemingly suggest that the opposite should be true. The greater concentration of companies in the Dow could be why there are significant results for both types of interest rates. In any case, it is clear that there is a relationship between
63 th e S&P and loan rate shocks, the Dow and loan rate shocks, and the Dow and FFR shocks. Starting with the S&P, a unit root was found in the data. It was logged to account for this and subsequently the growth rate of the S&P was found. The ARDL (3,3) in terms of the loan rate demonstrated significance in the first lag of S&P growth, the contemporaneous shock or the impact coefficient, and the first lag of shocks in the loan rate. A Wald F test was done to restrict the model, which was successful in limiting th e long term coefficient calculation to just the significant coefficients as the null hypothesis could not be rejected. This resulted in a positive value. The FFR regression did not yield similar results for the S&P. There was seemingly significance in the first S&P growth lag and the second FFR lag. However, the Wald F test was unsuccessful at restricting the model, and no long term coefficient was calculated. The weakness of this result may be because the S&P is an aggregate of so many stocks, and the amou nt of companies that are a part of this index made a significant result hard to find. In terms of the Dow, a unit root was also found, resulting in a variable that represents the growth the Dow. The ARDL (3,3 ) in terms of the loan rate demonstrates significance in the first lag of the Dow growth, the impact coefficient, and the first lag of the shock. The model was restricted successfully using the Wald F test, so the long term coefficient could be calculated. The same results were found for the shocks in the FFR. The positive long term coefficients found show that shocks in bank liquidity preference have a clear and significant relationship with multiple stock market variables, suggesting that they help drive fluctuations in the market. vi. Housing Starts
64 The hypothesis from chapter 3 held that a positive shock in bank liquidity preference would negatively affect the number of new homes being built. If banks decreased their liquidity preference, this would likely be associated with more new homes being built, as more money is coming out of reserves to lend to home builders and home buyers. Thus, there would be a negative relationship between housing starts and shocks in bank liquidity preference with the loan rate likely having the bigger impact. The hypothesis was supported by the long term coefficients of both the loan rate and the FFR. However, the liquidity preference shocks in terms of the FFR were larger than the loan rate shocks. A unit root was found in the housing start data, so it was logged to ensure it was stationary, creating the variable for housing start growth. When the initial ARDL (3,3) was performed for the loan rate, there was seemingly significance in the first and second housing starts lags, th e impact coefficient, and the second lag of the loan rate shocks. A Wald F test was performed allowing the model to be restricted to these significant coefficients from which the long term coefficient was calculated. When the ARDL (3,3) was calculated for the FFR, similar results were found. The first housing starts lag was statistically significant, as was the impact coefficient, and the second FFR shock lag. A Wald F test was also performed which was successful in restricting the long term coefficient cal culation to those significant coefficients. II. Effect of Bank Liquidity Preference Shocks on Macroeconomic Variables Using Impact Coefficients While the long term coefficient is the most important in gauging the effect of shocks in bank liquidity preferenc e on macroeconomic variables, the impact coefficients, or contemporaneous shocks in bank liquidity preference, can also be helpful in assessing
65 the effect of bank behavior on the broader economy, especially when they demonstrate some significance. Recall t hat the impact coefficient is the of this equation which is of course, the ARDL general equation. The impact coefficient is important because it shows how much the average change in the dependent variable ( ) will be when the independent variable ( ) changes by one unit. While more specific than the long term coefficient, the impact coefficient gives another look into how liquidity preference of banks can c hange key indicators in the economy. Table 4.2: Impact Coefficients Macro variables Impact coefficient estimate using the loan rate Impact coefficient estimate using FFR Unemployment 0.056437 ; t stat = 1.560033, p value = 0.1215 0.054126 ; t stat = 1.654840, p value = 0.1006 GDP 0.060494 ; t stat = 0.308679, p value = 0.7581 0.070520 ; t stat = 0.533550, p value = 0.5947 CPI 0.081110 ; t stat = 0.369843, p value = 0.7122 0.138470 ; t stat = 0.803384, p value = 0.4234 PPI 0.620196 ; t stat = 0.834223, p value = 0.4059 0.519887 ; t stat = 0.811916, p value = 0.4185 Corporate profits after taxes 4.588182 ; t stat = 1.184542, p value = 0.2386 3.701359 ; t stat = 1.188143, p value = 0.2372 S&P 1.80215 ; t stat = 2.111661, p value = 0.0368 1.557290 ; t stat = 1.093062, p value = 0.2766 Dow 1.78478 ; t stat = 2.872078, p value = 0.0048 1.16302 ; t stat = 1.749400, p value = 0.0827
66 Housing starts 2.51086 ; t stat = 6.453640, p value = 0.0000 2.36482 ; t stat = 2.686651, p value = 0.0083 In regards to unemployment, the impact coefficient is actually negative, which goes against the intuition that there should be a positive relationship evident. However, the lack of statistical significance of the coefficients suggests t hat there is likely no immediate impact of shocks on unemployment. While the GDP impact coefficient for the loan rate is positive, the one for the FFR is negative, but these results are not such a big deal given that the numbers are relatively small. Addit ionally there is clearly no statistical significance evident. The lack of significance in the impact coefficients is also demonstrated for CPI, PPI, and corporate profits after taxes. There is statistical significance demonstrated in the S&P in terms of t he loan rate. Clearly this coefficient is negative, which differs from the positive long term coefficient found. This suggests that in the current quarter, the shock in liquidity preference by a bank can actually hurt the S&P. The same can be seen in both of the impact coefficients for the Dow. This is not entirely surprising because when banks increase their liquidity preference, the money is being held in banks and not flowing outwards necessarily, especially in the short run. This behavior seems to chang e a bit in the long run, and that may make more sense given the speculative nature in the stock market. As a positive liquidity preference shock occurs, the market is likely to react very quickly and sharply downwards. However, as concerns about the increa se in liquidity in banks subside over time, they are less likely to have a negative impact in the stock market. There is also statistical significance in the housing starts impact coefficients. These coefficients are negative, which goes along with the in tuitive explanation for the
67 negative sign, as well as the long term coefficient. These impact coefficients work to further suggest that bank liquidity preference has a negative relationship with this macroeconomic variable and that this bank behavior is in strumental in driving fluctuations in key indicators in the economy. III. Limitations of Work and Future Research These results demonstrate that there are some statistically and economically significant relationships between shocks in liquidity preference by banks considering two different interest rates and key macroeconomic variables in the economy. While this sugge sts that bank liquidity preference behavior can drive major indicators in the economy and thus the business cycle in some key ways, the number of variables considered is clearly fairly small. Clearly more macroeconomic variables can be considered in the mo del to gauge how far and how deep shocks in bank liquidity preference affect changes in the economy. Considering more dependent variables would hopefully reveal a more robust relationship between bank liquidity preference shocks and the business cycle. Fo r example, some variables that could be added to the model include changes in commodity prices like gold, copper, and oil, among others. Changes in commodity prices have been very topical recently as commodity prices have fluctuated considerably since the financial crisis. Commodity prices have a great effect in the economy as their fluctuations can have a big effect on both demand and supply. Their effect can easily be seen when a severe increase in the price of oil affects the price of gas, which has nega tive effects on both the demand and supply sides of the economy. If bank liquidity preference is instrumental in the up or down movements of these prices that could be another big
68 indicator in showing that bank behavior drives changes in the economy. Given the speculative nature of the commodity market, it seems that an increase in liquidity preference could lead to an increase in the price of gold, for instance. These would be interesting relationships to study as well. Additionally, the use of really broa d aggregates to consider the macroeconomic variables was a good start, but more can be done to examine the complexities of these relationships. For instance while the GDP regressions did not yield significant results, maybe considering consumption as a sin gle macroeconomic variable and investment as another would demonstrate more significance. There seems to be something to the GDP and FFR shocks regression, but it could not be figured out using this methodology, so maybe something could be done to explore this relationship further, as well as the relationship between the inflation variables and FFR shocks. In regards to the corporate profits regressions, considering different sizes of corporations might be helpful as well. The variable used in the thesis e ncompasses a wide range of corporations, some of which have had much easier access to credit based on size and historical record. This may be very different than a smaller corporation, meaning that positive shocks in liquidity preference may be more harmfu l for a smaller an d less established corporation. Additionally, while bank liquidity preference has increased considerably as a result of the recession, so has corporate liquidity preference. Corporations seem to be sitting on more cash than ever as well, so seeing how this relates to the economy might also be an interesting topic to study. Furthermore, something that is emphasized throughout the paper is the connection between liquidity preference by banks and the credit market. As banks hold more money,
69 it is presumed that they are not lending it when they could be. On the flip side, as banks hold less money in reserves, they are lending more of it. This relationship is important in the thesis, but the mechanisms that allow this relationship to hold are n ever formalized. This might be an important issue to consider and work on as bank liquidity preference becomes more important as banks hold more money and credit accessibility continues to be a big topic.
70 Conclusion The financial crisis and subsequent global recession have provided many opportunities for meaningful research. Given that the Great Recession originated in the financial sector and particularly with the near death of all of the major commercial banks, the idea for this thesis came from exploring how the banks are able to affect the economy to such a great extent, as evidenced by the fact that the recession has affected the lives of virtually every American. In the wake of the recession, banks greatly increa sed their total reserves as the Fed has attempted to stimulate the economy through increasing the money supply. As this has occurred, there has seemingly been a very negative impact on the economy, especially as it has become more difficult to get lines of credit from these commercial banks. Thus this thesis has attempted to assess what happens in the economy when banks change how much money they are holding in reserves. I f this relationship is significant at all the ideas in this thesis would be an additi on to the literature. In order to demonstrate the lack of research and theory on the effects of financial intermediaries on the economy, a literature review was done considering all of the major theories in modern macroeconomics in regards to the busines s cycle. While notions of individual liquidity preference and frictions in the credit market were considered by some schools of thought, there is no real mention of the effect of bank liquidity preference on the economy. The second chapter considers the s ignificance of the idea of bank liquidity preference as it is related to financial liberalization, the recent global recession, and excess liquidity. It establishes that bank liquidity preference has become significant
71 because of these factors. Ultimately this chapter demonstrates that bank behavior matters, especially if it is keeping money out of the broader economy. The third chapter sets up the theoretical model on which these ideas were tested. First the shocks in bank liquidity preference are identi fied by considering that the amount of total reserves that banks hold are contingent upon different interest rates, but most importantly the bank prime loan rate and the federal funds rate, as these are basic interest rates in the economy. Both the intuiti on behind the idea and the empirical data demonstrate a downward sloping bank liquidity preference curve. Shocks in bank liquidity preference are evidenced by shifts in this curve. The data on key macroeconomic indicators give a sense of the changes happen ing in the broader economy in the past thirty years. An ARDL model was then considered for each macroeconomic variable, which gauged the effect of the shocks on each variable over time. Ultimately the fourth chapter reveals that significant relationships seem to exist between shocks in bank liquidity preference and real economic indicators, suggesting that bank behavior in this way helps drive the business cycle. While the model has its limitations, it seems that this thesis has hit on some key points tha t might not have been evident otherwise.
72 References Akerlof, George A., and Robert J. Shiller. Animal Spirits: How Human Psychology Drives the Economy, and Why It Matters For Global Capitalism Princeton University Press, 2010. Akitoby, Bernardin, Benedict J. Clements, Sanjeev Gupta, and Gabriela Inchauste. 2004. Term Behavior of Government Expenditures in IMF Working Paper Working Paper No. 04/202 : 1 24. Asteriou, Dimitrios, and Stephen G. Hall. Applied Econometrics: a Modern Approach Using EViews and Microfit New York, NY: Palgrave Macmillan, 2007. Baddeley, Michelle, and Diana Barrowclough. Running Regressions: a Practical Guide to Quantitative Research in Economics, Finance and Developm ent Studies New York: Cambridge University Press, 2009. Ball, Laurence. 2008. Money, Banking, and Financial Markets. Worth Publishers. The American Economic Rev iew 79(1): 14 31. Dow Jones Averages. Dow Jones Indexes. http://www.djindexes.com/averages/ (accessed April 10, 2012). Eggertsson, Gauti, and Jonathan D. Ostry. IMF Policy Discussion Paper 05/5 : 1 26 Federal Reserve Economic Data. FRED. http://research.stlouisfed.org/fred2/ (accessed April 10, 2012). Journal of Development Studies 41(4): 542 557. Friedm Review of Economics and Statistics 45(1): 32 64. Frost, Peter A. 1971. "Banks' Demand for Excess Reserves." The Journal of Political Economy 79(4): 805 825.
73 Federal Reserve Bank of St. Louis Review 91(2): 49 59. Hazard in Banking, and Pruden tial Regulation: Are Capital Requirements The American Economic Review 90(1): 147 165. A Study of U.S. Monetary Policy Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board, Washington, D.C. Staff Reports 380 Federal Reserve Bank of New Yo rk. Khemraj, Tarron. 2010."What does excess bank liquidty say about the loan market in Less Developed Countries?" Oxford Economic Papers 62(1): 86 113. Khemraj, Tarron, and Policy with a Lower The New School for Social Research Working Paper 04/2011 : 1 20. Koo, Richard C. 2009. The Holy Grail of Macroeconomics: Lessons from Japan's Great Recession Singapore: Wiley. Krugman, Paul, and Robin Wells. 2007. Economics New York: Worth Publishers. Journal of Japanese and International Economies 14: 221 237. Lindley, James T, Clifford B. Sowell, and Jr. WM. Stewart Mounts. 2001."Excess Reserves during the 1930s: Empir ical Estimates of the Costs of Converting Unintended Cash Inventory into Income Producing Assets." Journal of Economics and Finance 25(2): 135 148. The Journal of Economic Pers pectives 3(3): 79 90. Morrison, George. Liquidity Preference of Commercial Banks. Chicago: University of Chicago Press, 1966.
74 S&P 500. S&P Indices. http://www.standardandpoors.com/indices/ (accessed April 10, 2012). Saxegaard evidence from Sub IMF Working Paper 06/115 : 1 52. Schettkat, Ronald, and Jochem Langkau. Economic policy proposals for Germany and Europe Abingdon: Routledge, 2 008. Snowdon, Brian, and Howard R. Vane. Modern Macroeconomics: Its Origins, Development and Current State Northhampton, MA: Edward Elgar Pub, 2005. The Journal of Econ omic Perspectives 7(1): 45 65. Journal of SocioEconomics 3 3: 527 546.