p. 1

keynes savings paradox fisher s debt deflation and the banking crisis paul de grauwe university of leuven abstract the sharp fall in economic activity in the world is the result of an interaction between stock and flow deflation spirals these deflationary spirals have the same origin i.e a collective movement of fear and risk aversion animal spirits these lead economic agents savers firms and banks to take actions that create negative externalities making these actions self-defeating individuals savers firms banks are unable to internalize these externalities because collective action is costly we use a simple is-lm model to analyze the interactions between these different deflationary spirals we find that it is the interaction between the flow and stock spirals that create an unstable economy the banking crisis is at the center of this vicious downward spiral in order to solve the coordination failure implicit in the deflationary spirals the government must take action we describe the nature of the collective action by the government key is the resolution of the banking crisis without which the economy cannot be stabilized modern macroeconomic models based on the paradigm of the rational representative agent who understands the complexities of the world has become a misleading tool of analysis the problems relating to coordination failures and movements of collective fears that are at the core of the present macroeconomic reality play no role in these models it will not be surprising that these macroeconomic models have not informed us correctly about the nature of the economic crisis march 2009

[close]

p. 2

1 introduction the world economy is experiencing a downward spiral in output and international trade that has not been seen since the great depression of the 1930s the most striking feature of the downward spiral in economic activity is the speed at which it is occurring we show this in figure 1 the acceleration in the downturn of industrial production world trade and gdp since the end of 2008 is breathtaking figure 1 world industrial production imports and gdp-growth source ecb monthly bulletin march 2009 2

[close]

p. 3

how can such a rapid deceleration of economic activity be explained in this paper we identify different deflationary spirals these spirals feed on themselves and amplify each other we argue that this interaction of different deflationary spirals is at the core of the sudden deceleration of world economic activity we conclude with some ideas about how to stop these downward spirals 2 four deflationary spirals we identify four deflationary spirals we will argue that they have similar structures arising from coordination failures which in turn are triggered by animal spirits the four deflationary spirals are the keynesian savings paradox fisher s debt deflation the cost cutting deflation the bank credit deflation each of these deflationary spirals can be dealt with when they occur in isolation they become lethal when they interact with each other 2.1 keynesian savings paradox when one individual desires to save more and he is alone to do so his decision to save more consume less will not affect aggregate output he will succeed to save more and once he has achieved his desired level of savings he stops trying to save more when the desire to save more is the result of a collective lack of confidence animal spirits the individual tries to build up savings when all the others do the same as a result output and income decline and the individual fails in his attempt to increase savings he will try again thereby intensifying the decline in output and failing again to build-up savings there is thus a coordination failure if the individuals could be 3

[close]

p. 4

convinced that their attempts to build up savings will not work when they all try to do it at the same time they would stop trying thereby stopping the downward spiral somebody must organize the collective action an individual agent will not do this because the cost of collective action exceeds his private gain 2.2 fisher s debt deflation when one individual tries to reduce his debt and he is alone to do so this attempt will generally succeed the reason is that his sales of assets to reduce his debt will not be felt by the others and therefore will not affect the solvency of others the individual will succeed in reducing his debt when the desire to reduce debt is driven by a collective movement of distrust the simultaneous action of individuals to reduce their debt is self-defeating they all sell assets at the same time thereby reducing the value of these assets this leads to a deterioration of the solvency of everybody else thereby forcing everybody to increase their attempts at reducing their debt by selling assets here also there is a coordination failure if individuals could be convinced that their attempts to reduce their debt will no work when they all try to do this at the same time they would stop trying and the deflationary cycle would also stop an individual however will have no incentive to organize such a collective action 2.3 cost cutting deflation when one individual firm reduces its costs by reducing wages and firing workers in order to improve its profits and this firm is alone to do so it will generally succeed in improving its profits the reason is that the cost cutting by an individual firm does not affect the other firms the latter will not react by reducing their wages and firing their workers when cost cutting is inspired by a collective movement of fear about future profitability the simultaneous cost cutting will not restore profitability the reason is 4

[close]

p. 5

that the workers who earn lower wages and the unemployed workers who have less or no disposable income will reduce their consumption and thus the output of all firms this reduces profits of all firms they will then continue to cut costs leading to further reductions of output and profits there is again a coordination failure if firms could be convinced that the collective cost cutting will not improve profits they would stop cutting their costs but individual firms have no incentives to do this 2.4 bank credit deflation when one individual bank wants to reduce the riskiness of its loan portfolio it will cut back on loans and accumulate liquid assets when the bank is alone to do so and provided it is not too big it will succeed in reducing the riskiness of its loan portfolio the reason is that the strategy of the bank will not be felt by the other banks which will not react once the bank has succeeded in reducing the riskiness of its loan portfolio it will stop calling back loans when banks are gripped by pessimism and extreme risk aversion the simultaneous reduction of bank loans by all banks will not reduce the risk of the banks loan portfolio there are two reasons for this first banks lend to each other as a result when banks reduce their lending they reduce the funding of other banks the latter will be induced to reduce their lending and thus the funding of other banks second when one bank cuts back its loans firms get into trouble some of these firms buy goods and services from other firms as a result these other firms also get into trouble and fail to repay their debt to other banks the latter will see that their loan portfolio has become riskier they will in turn reduce credit thereby increasing the riskiness of the loan portfolio of other banks there is again a coordination failure if banks could be convinced that the simultaneous loan cutting increases the risk of their loan portfolio they would stop cutting back on their loans and the negative spiral would stop they have no individual incentives however to engage in collective action 5

[close]

p. 6

the four deflationary spirals that we described have the same structure the action by one economic agent creates a negative externality that makes this action self-defeating this spiral is triggered by a collective movement of fear distrust or risk aversion animal spirits individuals savers firms banks are unable to internalize these externalities because collective action is costly there is thus a failure to coordinate individual actions to avoid a bad outcome it is important to stress that these market failures are triggered by animal spirits i.e waves of correlated beliefs for a fascinating recent analysis see akerlof and shiller2009 when these beliefs are not correlated the market will work fine to give an example when an individual firm is alone to believe it will have a profit problem in the future it will cut costs if other firms have no such beliefs they will not cut their costs this will allow the first firm to improve its profitability the market system will work fine in coordinating the different uncorrelated beliefs of individual firms when the beliefs are correlated i.e when animal spirits occur the market will fail to coordinate individuals actions towards a good equilibrium 1 cyclical movements in optimism and pessimism animal spirits have always existed why do these now lead to such a breakdown of coordination we answer this question in the next section 3 flow and stock deflations the four deflationary spirals we identified although similar in structure are different in one particular dimension numbers one and three savings paradox and cost deflation can be called flow deflations they arise because consumers and firms want to change a flow savings and profits numbers two and four involve the adjustment of stocks the debt levels and the levels of credit we call them stock deflations problems arise when the flow and stock deflations interact with each other note that ultimately these market failures occur because of an information problem agents experience cognitive problems in understanding the world as a result they use limited information and tend to follow the lead of others this will sometimes lead to correlated beliefs that are self-fulfilling and self-reinforcing 1 6

[close]

p. 7

in normal recessions such as the ones we have experienced in the postwar period prior to the present crisis only the flow deflations were in operation there had not been a preceding period of excessive debt accumulation and unsustainable levels of bank loans as a result households firms and banks were not trying to adjust their balance sheets the pessimism of households and firms was related to expected shortfalls in income and profits and led to increased savings and cost cutting in such an environment in which the stock levels were perceived to be right there were sufficient automatic equilibrating mechanisms that prevented these two flow deflations from leading to an unstoppable downward spiral the most important equilibrating mechanism occurred through the banking system when banks function normally they have a stabilizing force on the business cycle the reason is that in a recession the central bank typically reduces the interest rate making it easier for banks to lend in normal circumstances when banks are not in the process of cleaning up their balance sheets they will be willing to transmit this interest rate decline into a reduction of the loan rate as a result banks will engage in be automatic distress lending to firms and households households will be less tempted to increase their savings in addition private investment by firms will be stimulated i.e firms will be willing to dissave thereby mitigating the deflationary potential provoked by the savings paradox in section 5 we show this in the context of a simple is-lm analysis the interest rate decline will also mitigate the cost cutting dynamics this is so because it improves the profit outlook for firms giving them lower incentives to go on cutting costs thus when the banking system functions normally there are self-equilibrating mechanisms that prevent the flow deflations from degenerating into uncontrollable downward spirals the problem the world economy faces today is that flow and stock deflations interact and reinforce each other the period prior to the crisis was one of excessive buildups of private debt and banks assets the result of these excessive buildups of private debt and balance sheets is that the stock deflation processes described in the previous section operate with full force as a result the equilibrating mechanism that exists in normal recessions does not function the lower interest rates engineered by central banks are not 7

[close]

p. 8

transmitted by the banking sector into lower loan rates for consumers and firms in addition we now are confronted by the interaction of the flow and stock deflations this interaction amplifies these deflationary processes in section 5 we show such an interaction between the keynesian savings paradox fisher s debt deflation and the banking crisis this interaction can be described as follows because of excessive debt accumulation of the past households desire to reduce their debt levels thus they all attempt to save more as argued earlier these attempts are self-defeating as a result households fail to save more and thus fail to reduce their debt this leads them to increase their attempts to save more the fact that the banks do not pass on the lower deposit rates into lower loan rates makes things worse there are no incentives for firms to increase their investments no dissaving nothing stops the deflationary spiral the interaction goes further the deteriorating conditions in the real economy feed back on the banking system banks loan portfolios deteriorate further as a result of increasing default rates banks reduce their lending even further etc 4 intermezzo why has modern macroeconomics become irrelevant modern macroeconomics as embodied in dynamic stochastic general equilibrium models dsge is based on the paradigm of the utility maximizing individual agent who understands the full complexity of the world since all individuals understand the same truth modern macroeconomics has taken the view that it suffices to model one representative individual to fully represent reality thus as a consumer the agent continuously maximizes an intertemporal utility function and is capable of computing the implications of exogenous shocks on his optimal consumption plan taking full account of what these shocks will do to the plans of the producers similarly producers compute the implications of these shocks on their present and future production plans taking into account how consumers react to these shocks thus in such a model coordination failures cannot arise the representative agent fully internalizes the external effects of all his actions when shocks occur there can be only one equilibrium to which the system will converge and agents perfectly understand this 8

[close]

p. 9

in such a world animal spirits i.e independent waves of optimism and pessimism cannot arise the absence of animal spirits and of coordination failures makes sure that deflationary spirals as we have described them in the previous sections cannot occur in the world of the dsge-models the macroeconomic upheavals that the world economy is experiencing cannot be denied even by dsge-modelers the way the latter treat these upheavals is to introduce them as exogenous shocks see smets and wouters2007 woodford2009 thus the credit crisis that erupted in august 2007 is the result of an exogenous increase in risk aversion and not the result of previous periods of euphoria that led to bubbles in asset markets and unsustainable debt levels in the dsge-world the crisis is the result of an event that could not be foreseen like a meteor hit this shock forces rational agents to adjust their consumption and production plans once they will have done this we will be back in the best of all possible worlds it will not come as a surprise that models that are populated by rational agents who understand the full complexities of the world and who all agree on how to interpret these complexities have not produced any useful insight allowing us to understand the nature of the present economic crisis yet vast amounts of intellectual energies are still being spent on the further refining of dsge-models 9

[close]

p. 10

5 keynesian savings paradox fisher debt deflation and banking crisis in a simple is-lm framework in this section we analyze more formally the interaction between three of the four deflation processes we have identified i.e the savings paradox the debt deflation and the banking crisis the keynesian savings paradox can be represented in the context of the traditional islm model we assume that the economy is in a liquidity trap there are different versions of the liquidity trap here we assume the modern version i.e the commercial banks have unknown amounts of impaired assets on their balance sheets as a result they have stopped providing new lending and use the central bank s liquidity injections as a means to reduce the riskiness of their assets thus central bank liquidity is hoarded by the banks and does not lead to a lower interest rate for consumers and firms the economy operates in the horizontal segment of the lm-curve let us now assume that consumers desire to increase their savings a flow deflation this shifts the is curve to the left from is1 to is2 in figure 1 in the lower part of figure 1 we also represent the balance between investment and savings we assume that savings are a function of income this relation is represented by the s-curve investment is a function of the interest rate and is represented by the ir line the desire to save more leads to a shift of the s-line from s1 to s2 the keynesian savings paradox can immediately be derived from figure 1 the attempted increase in savings the shift of the s-curve fails to lead to more savings the reason is that the ensuing decline in income y lowers the volume of savings consumers have less income to save from in the new equilibrium b and b effective savings are exactly equal to the initial savings since this is a pure flow shock we reach a new equilibrium in b and b consumers stop attempting to save more the new equilibrium is stable although unpleasant since output has declined 10

[close]

p. 11

figure 1 interaction of flow and stock deflation with an impaired banking system r is shifts to the left because of increase in desired savings is1 lm is2 is3 cbayc increase in desired savings but actual savings do not s increase b a ir s1 s3 s2 things get more interesting for economists at least not for the people involved if we allow for an interaction between this keynesian flow analysis with the stock analysis of irving fisher see fisher1933 suppose that the reason why consumers attempt to increase their savings is that they have accumulated too much debt in the past they now want to reduce the level of their debt a fixed number by increasing their savings what does this imply in the analysis of figure 1 as before we have a shift of the is curve from is1 to is2 and a shift of the s-curve from s1 to s2 in b and b consumers find out that they have not been able to save more as a result they have also failed to lower their debt levels which have remained unchanged they will therefore not stop their attempts 11

[close]

p. 12

to save more since this is the only way they can bring down their debt the s-curve shifts further downwards to s3 and the is-curve shifts further to the left to is3 we obtain a new equilibrium in c and c it can however immediately be seen that this is not a sustainable equilibrium because in c consumers still have not been able to increase their savings and thus their debt levels are still the same as in the initial point thus they will continue their attempts to increase savings that will continue to be self-defeating we have a true deflationary spiral that does not stop the economy is unstable from the preceding analysis we conclude that the interaction between the flow and stock deflations creates a very different dynamics than the simple keynesian flow analysis in the latter case the increase in savings leads to a decline in output but the economy settles in a lower but stable output equilibrium this is not the case when we allow for stock and flow interactions in this case we obtain a true deflationary spiral see also minsky1986 the previous analysis implicitly assumed an impaired banking system which leads to a liquidity trap a flat lm-curve thus we assumed a double stock problem i.e a debt problem of consumers and a banking problem what if we relax the banking problem we do this in figure 2 we now assume that the lm curve has the normal upward slope thus when the central bank lowers the interest rate this is passed on to consumers and firms as a result the economy is stimulated we analyze the effect of an increase in desired savings in the same way as in figure 1 the is and s-curves shift downwards to is2 and s2 respectively so that the interest rate declines from r1 to r2 this leads to an increase in investment this is shown in the lower part of figure 2 by the shift of ir1 to ir2 we now find that in the new equilibrium b and b savings have increased relative to the initial level in a thus consumers succeed in reducing their debt levels the fundamental reason is that firms have decided to invest more i.e to dissave in other words somebody must be willing to dissave for consumers to succeed in saving more and to reduce their debt levels we now obtain an important result suppose the economy stays in the new equilibrium b and b this means that at the end of the next period consumers will have added additional savings which allows them to reduce their debt levels even further thus at the end of that period debt levels will be lower than at the end of the previous period this 12

[close]

p. 13

will go on this implies that just staying in the equilibrium b and b will lead to a continuous decline in debt levels until consumers have reached their desired debt levels they will stop saving more at that moment they may decide to go back to their initial desired savings as a result the s-curve will shift back up again leading to a stimulus in economic activity figure 2 interaction of flow and stock deflation with well-functioning banking system is1 r is shifts to the left because of increase in desired savings is2 a lm r1 r2 b y a increase in desired savings actual savings increase partially b ir1 ir2 s1 s2 s 13

[close]

p. 14

we conclude that fixing the banking system allows the economy to find a stable equilibrium even if the keynesian flow deflation interacts with the fisher debt deflation it should be noted that even in this case the decline in output may still be uncomfortably high and may last for some time nevertheless the result is qualitatively different in that there is now a stable equilibrium this contrasts with the case when the keynesian flow deflation interacts with the fisher debt deflation and with a banking crisis thus the key problem is the banking crisis whish prevents the system from finding a stable equilibrium 6 role of government solving the coordination failures by taking collective action we have argued that the intensity of the present economic downturn is the result of the fact that the flow deflations that are characteristic of normal recessions coincide and interact with stock deflations the latter are the result of previous unsustainable debt and asset price developments that now have to be wound down the common characteristic of these different deflationary spirals is a coordination failure the market fails to coordinate private actions towards an attractive collective outcome this market failure can in principle be solved by collective action such a collective action can only be organized by the government let us analyze what this collective action must be to deal with the different forms of deflation as will be clear the principles are easy to formulate the implementation is more difficult the key to economic recovery is the stabilization of the banking sector the simple model of section 5 teaches us that a banking sector that is in the grips of credit deflation and deleveraging can destabilize the economy and can push the economy into a true deflationary spiral there is no secret about how the bank credit deflation can be stopped here are the principles see hall and woodward2009 for more detailed analysis first bad loans should be separated from good loans putting the former in separate entities bad banks to be managed by specialized management teams whose responsibility it is to dispose of these assets losses on these bad assets are inevitable and so is the inevitability that the 14

[close]

p. 15

taxpayer will be asked to foot the bill the managers of the bad bank should minimize these losses the good loans remain on the balance sheet of the good bank the hope is that this good bank freed as it is from the toxic assets will feel liberated and will be willing to take more risk so that the credit flow can start again one can doubt however that a privately run good bank will have sufficient incentives to start lending again the reason is that extreme risk aversion and a desire to save the skin of the shareholders will restrain the managers of the good bank in extending loans if that is what the managers of the good bank do the bank credit deflation process described earlier will not stop this leads to the issue of whether it is not desirable to temporarily nationalize the good bank such nationalization would take away the paralyzing fear that new bank loans put the bank s capital and its shareholders at risk there is a second reason why the government may want to temporarily nationalize the good bank the bad bank ­ good bank solution carries the risk of socializing the losses while privatizing the profits indeed the losses of the bad bank will necessarily be borne by the taxpayers if then the good bank remains in private ownership the expected future profits will be handed out to the shareholders but these profits will be realized only because the toxic assets have been separated and the losses on these assets have been borne by taxpayers it is therefore more reasonable to make sure that these future profits are reserved for compensation of the taxpayers who have paid for the losses on the bad assets the resolution of the bank crisis along the lines discussed in the previous paragraphs is a necessary condition for the recovery it will also make the use of other macroeconomic policies easier and more effective these other macroeconomic policies must be geared towards resolving the other deflationary processes let us discuss these consecutively 15

[close]