«What do ®nancial intermediaries do? a,* b Franklin Allen, Anthony M. Santomero a The Wharton School, University of Pennsylvania, Room 2336, ...»
286 F. Allen, A.M. Santomero / Journal of Banking & Finance 25 (2001) 271±294 and Japan are broadly similar. To normalize for the size of each countryÕs GDP, Fig. 16(b) reports ®nancial assets as a percentage of GDP. France and Germany have a signi®cantly lower amount of ®nancial assets with ratios less than 2 for Germany and 1.5 for France. Combining the results illustrated in Figs. 15 and 16 shows that taking into account the amount of wealth held in ®nancial assets increases the dierences in the amount of risk borne by households in the dierent countries, rather than reduces it. Not only do households hold much higher proportions in risky securities in the US and UK, they also hold more ®nancial assets, particularly relative to France and Germany.
How can one explain these dierences in the amount of risk households are apparently exposed to in dierent ®nancial systems? Standard ®nancial theory suggests that the main purpose of ®nancial markets is to improve risk sharing.
Financial markets in the US and UK are more developed by most measures than in Japan and France and much more developed than in Germany. How can it be that households are exposed to more risk in the US and UK than in Japan, France and Germany?
Allen and Gale (1997, 1999a) have provided a resolution to this paradox.
They point out that traditional ®nancial theory has little to say about hedging nondiversi®able risks. It assumes that the set of assets is given and theory focuses on the ecient sharing of these risks through exchange. For example, the standard diversi®cation argument requires individuals to exchange assets so that each investor holds a relatively small amount of any one risk. Risks will also be traded so that more risk-averse people bear less risk than people who are less risk-averse. This kind of risk sharing is termed cross-sectional risk sharing, because it is achieved through exchanges of risk among individuals at a given point in time. However, importantly, these strategies do not eliminate macroeconomic shocks, which aect all assets in a similar way.
Departing from the traditional approach, Allen and Gale focus on the intertemporal smoothing of risks that cannot be diversi®ed at a given point in time. They argue that such risks can be averaged over time in a way that reduces their impact on individual welfare. One hedging strategy for nondiversi®able risks is intergenerational risk sharing. This spreads the risks associated with a given stock of assets across generations with heterogeneous experiences.
Another strategy involves asset accumulation in order to reduce ¯uctuations in consumption over time. Both are examples of the intertemporal smoothing of asset returns.
Allen and Gale show that the opportunities for engaging in intertemporal smoothing are very dierent in market-based and bank-based ®nancial systems. They demonstrate that incomplete ®nancial markets may not allow effective intertemporal smoothing, but long-lived ®nancial institutions, such as banks, can allow this, as long they as are not subject to substantial competition from ®nancial markets. In fact, such competition can lead to the unraveling of F. Allen, A.M. Santomero / Journal of Banking & Finance 25 (2001) 271±294 287 intertemporal smoothing provided by long-lived institutions. This is because in good times individuals would rather opt out of the banking system and invest in the market, thus avoiding the accumulation of reserves which they may not bene®t from. Therefore, in the long run intertemporal smoothing by banks is not viable in the presence of direct competition from markets. 6 This theory provides a framework for thinking about the role of risk management in dierent ®nancial systems. In bank-based systems such as those in Japan, France and Germany, risk management can be achieved through intertemporal smoothing. Financial intermediaries eliminate risk by investing in short term liquid assets. Other kinds of risk management are relatively less important since cross-sectional risk sharing is correspondingly reduced in importance.
On the other hand, in market-based ®nancial systems intertemporal smoothing by intermediaries is ruled out by competition from ®nancial markets. Here, cross-sectional risk sharing becomes correspondingly more important. As a result individuals or institutions acting on their behalf, need to trade and manage risk in a very dierent way. They need to ensure that those who are most tolerant of risk end up bearing it. The Allen and Gale theory thus predicts that as ®nancial systems become more market oriented, risk management through the use of derivatives and other similar techniques will become more important. The theory is thus consistent with the fact that risk management is much more important in the US and UK than it is in less market oriented economies such as Japan, France and Germany.
It is also consistent with the rapid increase in the amount of risk management that has been undertaken in recent years. It can be seen from Fig. 2 that the US ®nancial system has changed dramatically in the last century. Until the middle of the 1970s, banks were the dominant form of ®nancial institution. The structure of the ®nancial system was much more like that of Japan, France and Germany, in the sense that banks dominated and markets were less signi®cant.
It can be argued that up until then banks were able to engage in intertemporal smoothing. However, the ®nancial innovation that has occurred in the last 25 years changed the form of the ®nancial system from a bank-based one to a market-based one. As the previous sections demonstrated, banks in the US have increasingly had to compete with markets. This has meant that risk management using derivatives markets and so forth has become increasingly important. Thus, the amount US intermediaries have engaged in this type of risk management has increased signi®cantly, because the amount of risk management through intertemporal smoothing has necessarily been reduced.
6 See Petersen and Rajan (1995) for a somewhat dierent model that investigates smoothing from the lending side.
288 F. Allen, A.M. Santomero / Journal of Banking & Finance 25 (2001) 271±294
4. A dynamic theory of intermediation A complete theory of intermediation should explain both static and dynamic aspects of the process. In short it should develop dynamic theories which explain both the need and reality of ®nancial innovation. Here, we give an explanation of the rationale behind these dynamic elements and pressures that favor such innovation without developing a complete theory.
As the analysis in the previous two sections has indicated, we believe that there has been a fundamental shift in the nature of intermediation in the US and also the UK. The amount of risk borne by investors in these countries is dierent than once was the case, and is also higher than the amount currently borne by households in Japan, France and Germany.
In a traditional bank-based economy, where ®nancial markets are not very signi®cant, the main way in which banks deal with risk is through intertemporal smoothing. They acquire a ``buer'' of short-term liquid assets when times are good and run this buer down when times are bad. As a result of this buer, households that hold most of their assets in bank accounts and other ®xed income assets are to a large extent shielded from risk and are able to have smooth consumption streams.
When ®nancial markets develop they provide competition to banks which makes the intertemporal smoothing they undertake increasingly more dicult.
Financial markets allow high returns in good times and there is an incentive for individual investors to withdraw their funds from banks and put them in markets instead. In order to survive the banks must essentially compete with the markets and cease oering intertemporal smoothing.
The data we presented in Section 2 showed how this fundamental change has altered the business that banks undertake in the US. As we documented there, banks have been forced to be more entrepreneurial and to innovate in order to survive. They have entered new markets and developed new products.
Their traditional role of taking deposits and making loans has been steadily shrinking. Despite regulation designed to ensure they limit their activities to commercial banking, they have been able to increasingly skirt these regulations and lobby for changes. This has allowed them to enter the underwriting, distribution, asset management, and insurance businesses, as well as develop new products. 7 They have succeeded in replacing their traditional business with fee-producing activities, as Fig. 10 illustrated. This has been accomplished by institutions shifting their role from principal to agent in many transactions.
This move is quite consistent with the shift from intertemporal smoothing to cross-sectional risk sharing outlined in Section 3. The role of surviving intermediaries has become one in which they have increasingly facilitated risk 7 See Santomero and Babbel (1997), Chapter 24, for a detailed discussion of this process.
F. Allen, A.M. Santomero / Journal of Banking & Finance 25 (2001) 271±294 289 transfer and allocate it to those most able to bear it. Achieving this new role requires considerable change and innovation on the part of individual institutions and individual bankers as well.
In recent years a considerable literature on ®nancial innovation has developed (see, e.g., Allen and Gale (1994) and Due and Rahi (1995) for overviews). However, this literature has been concerned mainly with the innovation of securities in ®nancial markets, rather than understanding the role of intermediaries in the process, or the dynamics of change within the institutions themselves. More needs to be done to better understand this challenge.
In AS we suggested that participation costs are important to understanding modern intermediaries and their new role. We believe that lowering participation costs would appear to be an important motivation for much of the innovation that has taken place in ®nancial services. It is simply too costly for individuals to directly manage risk using modern techniques.
A key point to appreciate, however, is the precise meaning of participation cost in this context. One interpretation of participation cost is that it is simply the time involved in making ®nancial decisions. AS point out that the opportunity cost of peopleÕs time, particularly for many professionals, has increased signi®cantly in the last 15 years. Scholtens and van Wensveen (1999) have argued that wages in fact have been ¯at in the US during this period.
While true in the aggregate, this is not the case for professionals who undertake the majority of savings, because income inequality has increased over this same period. More importantly, though, this is only part of the costs of participating in markets.
A much broader notion of participation costs is required to understand the kinds of innovation that are occurring. The degree of sophistication and specialization required to undertake complex risk trading and risk management operations is very high. It is the problems associated with acquiring and using this expertise that are really at the core of what we mean by participation costs.
Much of what modern intermediaries do is to interface between individuals and increasingly complex ®nancial markets. In order to do this eectively intermediaries must make their products relatively simple to understand, even if they are not simple to implement. Alternatively, they must develop a relationship based on trust with the investors they serve. Allen and Gale (1999b) have argued that the complex problems involved in delegating decisions to an intermediary (agent), when the investor (principal) does not fully understand the nature of the problem being solved, can be overcome with long term relationships. Financial markets and ®nancial intermediaries then have a symbiotic relationship. Each is necessary to the other. Without intermediaries, the informational barriers to participation would prevent investors from reaping the bene®ts of new markets, and the markets themselves might not survive. At the same time, ®nancial markets allow intermediaries to hedge cross-sectional risks more eectively than they previously could.
290 F. Allen, A.M. Santomero / Journal of Banking & Finance 25 (2001) 271±294 Viewed in this context innovation becomes central to the theory of intermediation. Banks and other intermediaries need to develop new ways to lower participation costs. This allows everybody to bene®t from improved crosssectional risk sharing. Merton and Bodie (1995) have termed this the ``®nancial innovation spiral''.
5. Concluding remarks
The world ®nancial system has changed signi®cantly in recent decades. In the US, banks and many other types of intermediaries have moved away from their traditional role of taking deposits and making loans. Although their share of intermediated funds has fallen they have not shrunk relative to GDP, and they remain an important part of the ®nancial system. They have achieved this by moving away from simple balance sheet intermediation toward feeproducing activities.
A comparison of the modern US ®nancial system with other ®nancial systems shows broadly similar trends but signi®cantly distinct operating models.
Broadly speaking, the US and UK appear most similar, with France and Japan moving toward the Anglo-Saxon model, and Germany still quite far away.
One result is that risk is borne dierently in these economies, as can be seen by the aggregate portfolio composition of households. This shows that US household allocations are similar to the UK but quite dierent from Japan, France and Germany. In the US and UK individuals hold signi®cantly more risky assets, and are exposed to more risk than in Japan, France and Germany.
We argued that this dierence could be understood using the framework of Allen and Gale (1997, 1999a). In the absence of signi®cant competition from ®nancial markets as in Japan, France and Germany, intermediaries are able to eliminate risk by intertemporal smoothing. They build up reserves of short term liquid assets when returns are high and run them down when they are low.