تقسیم بندی جغرافیایی بازار سرمایه ایالات متحده

Demographic variation in savings behavior can be exploited to provide evidence on segmentation in US bank loan markets. Cities with a large fraction of seniors have higher volumes of bank deposits. Since many banks rely heavily on deposit financing, this affects local loan supply and economic activity. I show a positive effect of local deposit supply on local outcomes, including the number of firms, the number of manufacturing firms, and the number of new firms started. The effect is stronger in industries that are heavily dependent on external finance. The deregulation of intrastate branching reduced the effect of local deposit supply by approximately a third.

In most economies, banks play a large role in the intermediation of capital from suppliers to users. Unlike financial markets, banks are principally local intermediaries. This applies to both sides of the balance sheet. On the liability side, banks rely heavily on deposits for funding (Kashyap and Stein, 2000) and most deposits are local. On the asset side, much bank lending is local (Petersen and Rajan, 2002). For these reasons, local variation in the supply of deposits could translate to local variation in the availability and cost of capital for borrowers and, hence, in the level of economic activity. Geographical segmentation is difficult to identify empirically, however. A direct approach is to examine the correlation between local lending volumes and local economic outcomes. In practice, this approach suffers from a severe endogeneity problem because the volume of local bank lending is likely to respond to both supply and demand for loans. The demand for loans is trivially correlated with economic outcomes. So, to evaluate whether the local supply of capital affects local economic outcomes, a source of exogenous variation in the supply of bank loans is necessary. This paper utilizes demographic variation in the supply of deposits as such a determinant of local capital supply. Seniors (those age 65 and older) tend to hold higher levels of bank deposits than other groups both in absolute terms and as a fraction of portfolios. However, because seniors do not participate much in the labor market or operate businesses, and because they consume less than other groups, the impact of a large fraction of seniors on the local demand for business finance is likely to be small and perhaps negative (See Section 2 for a more detailed discussion of the effect of seniors on loan demand.). Hence, a large fraction of seniors in an area causes a higher supply of intermediated finance relative to local demand for external financing. This makes the fraction of seniors a potentially useful instrument for the local supply of finance (relative to demand). Substantial demographic variation exists within the US, both across and within states. I use data on the fraction of seniors at the level of metropolitan statistical areas (MSAs) to predict deposit volumes and loan availability. This level of geographical detail permits including state fixed effects in regressions. Exploiting only within-state variation, I show that a high fraction of seniors corresponds to a high supply of deposits. In areas with high deposit supply, local banks use relatively more deposit financing (as opposed to equity and nondeposit debt) and have more liquid balance sheets (as measured by holdings of treasury securities).1 Using seniors as an instrument for deposit supply, I show that this supply is related to local economic outcomes. MSAs with high levels of deposits have more firms, more manufacturing firms and establishments, relatively more small firms (up to 19 employees) and fewer large firms (more than five hundred employees), and more new firm starts. Using a measure of dependence on external finance, based on Rajan and Zingales (1998), I show that the effect of the local deposit supply is stronger in industries that are more externally dependent (i.e., those industries in which, on average, Compustat firms use more external financing). I next examine the robustness of these results to several possible concerns. To address the potential endogeneity of seniors, demographic predictions are used in the place of actual data on seniors (middle-age people 20–30 years in advance), and the results remain similar and significant. I also consider the possibility that the seniors variable is correlated with wealth, perhaps driving economic outcomes through an effect on demand. I attempt to control for wealth by including average local house prices and per capita income as control variables. This does not affect coefficient estimates much. To examine whether the market definition used (MSAs) is too wide, results are presented for smaller areas (ZIP Codes). Smaller MSAs tend not to contain many ZIP Codes, so I focus on the three largest cities (New York, Los Angeles, Chicago). Within these cities, the ZIP Code-level deposit supply has a positive effect on the number of establishments. Finally, time-series tests, although exploiting limited variation in the data (because the fraction of seniors is stable), find a positive, borderline significant effect of deposits on outcomes. A final set of robustness tests addresses the issue of where integration is weakest, i.e., where local deposits have the strongest impact on local outcomes. There are multiple potential channels for geographical reallocation of savings, in either of the steps from depositor to intermediary and on to borrower. Furthermore, internal and external capital markets could allow funds to be transferred from bank branches in high deposit areas to branches with lower deposit supply (or higher loan demand). Frictions involving regulation, agency problems and information asymmetries could limit the scope for geographic transfers, however. In Section 6, evidence is presented showing that deregulation of intrastate branching (see Kroszner and Strahan, 1999) reduced the effects of geographical variation by approximately one third. This suggests that one of the important benefits of deregulation has been better geographical integration of capital markets. The increasing integration of local capital markets could be a process that is still going on. Geographic variation in the supply of capital in all likelihood implies welfare costs. If the marginal productivity of capital is declining, locations with a higher local supply of deposits that employ more capital use capital at lower marginal productivity than locations with less capital. Therefore, such variation reduces aggregate output compared with a frictionless world.2 The magnitude of welfare costs depends primarily on the rate at which capital productivity declines with capital intensity and could vary by sector and over time. A few caveats are in order. Bank lending data are not available by bank branch, so I cannot examine the effect of deposit supply on the pricing of local loans. Second, time-series variation is limited for the fraction of seniors, so time-series results must be considered provisional. This paper is related to papers showing that the supply of loans affects economic outcomes. Peek and Rosengren, 1997 and Peek and Rosengren, 2000 show that variation in Japanese banks’ lending in the US, induced by Japanese economic events, has had a large effect on construction activity in California, Illinois, and New York. Ashcraft (2005) shows that in two cases when the Federal Deposit Insurance Corporation closed healthy banks (to cover losses at affiliated banks), bank loans and local incomes declined. In contrast to these results, Driscoll (2004) uses estimated state-specific shocks to the demand for money to instrument for loan supply and finds a positive relation to the volume of bank lending, but no effect on output. The results in this paper complement and expand on these findings in several ways. The instrument is more generally available, not depending on infrequent regulatory action (such as the Texas bank closures) or specific foreign episodes (such as the Japanese boom). This offers opportunities for broader uses, e.g., comparisons over time or international comparisons of financial systems. Also, the instrument is available at fine geographical levels (e.g., MSAs). This paper is also related to research showing that there is substantial within-country variation in financial systems. Jayaratne and Strahan (1996) analyze variation in bank regulation across US states and find that deregulation of entry and mergers substantially increased output growth rates. Cetorelli and Strahan (2006) report that, at the state level, the number of firms is more sensitive to bank competition in industries with high dependence on external financing. With US data, Garmaise and Moskowitz (2006) use bank mergers as an instrument for local bank competition, and also find negative temporary effects on loan supply and economic activity following a bank merger. Using Italian data, Guiso et al. (2004) show that financial development (the probability that a household is shut off from the credit market) is largely determined by regulatory restrictions on bank entry. They find that financial development affects rates of new firm entry, growth of existing firms (especially small firms) and product market competition. My findings match this literature in suggesting that local credit markets matter for economic outcomes. However, the focus here is not on variation in local competitiveness and institutional quality, which affect local financial intermediation, but on geographical segmentation of capital markets, which affects the transfer of capital across space. The implications are therefore different. In particular, much of the literature has suggested that bank competition is beneficial and therefore implicitly or explicitly that mergers could be harmful (e.g. Sapienza, 2002 and Garmaise and Moskowitz, 2006). The evidence on segmentation in fact suggests that a cost of a dispersed banking system is that geographical segmentation is exacerbated. Thus, there are potential welfare benefits of mergers that combine banks from different areas.3 Finally, my findings on deregulation are related to the results of Jayaratne and Strahan (1996), who show large positive effects on state level growth rates after deregulation. Bank deregulation reduced the impact of local deposit supply, implying that geographical segmentation was reduced. This suggests a further channel through which the benefits identified could have come about. The rest of the paper is organized as follows. In Section 2, I show that the deposit holdings of seniors are higher, and their labor market participation lower, than that of other age groups. In Section 3, I discuss the conditions under which local deposits are likely to affect bank lending and those under which bank lending in turn is likely to affect economic activity. Section 3 covers the data used, and Section 4 the basic empirical results. In Section 5, I introduce additional control variables and robustness tests, and in Section 6 discuss what my results imply about the nature of financial frictions and presents some tests of frictions. Section 7 concludes.

I use the fraction of seniors as an instrument for the local supply of deposits and, hence, of bank loans. The supply of deposits affects economic outcomes: When deposits are high as a result of demographics, there are more firms, especially small, both overall and in manufacturing, as well as more new firms. The results are robust to several alternative specifications: using demographic predictions of future senior numbers, including controls for local wealth, and reducing area size from MSA to ZIP Code. Time-series evidence based on changes in seniors, though based on limited variation, also supports the main cross-sectional findings. If capital has a declining marginal productivity at the MSA level, the welfare implications of variations in the supply of capital to firms are negative. Capital is used at lower marginal productivity in locations with higher loan supply, whereas, in a frictionless world, this capital could be reallocated to a different location with higher marginal productivity of capital. Bank deregulation has been shown to have positive effects on economic outcomes (see, e.g., Jayaratne and Strahan, 1996 and Morgan et al., 2004). In terms of mechanisms, the focus has been on bank competition. My results suggest that the effect of bank deregulation on geographical allocation of capital could be important as well. I estimate that the effect of local deposit supply on the local economy is a third weaker after deregulation than before it. This suggests that US state-level deregulation reduced geographical segmentation (perhaps through increased bank size improving access to internal and external capital markets). This improved geographical allocation could have important welfare benefits. This conclusion has clear regulatory implications. A banking system with small and geographically fragmented institutions (as, e.g. Spain, the US and Italy have had historically) is inefficient because it aggravates the geographical segmentation of capital markets.25 This is likely to be especially severe in geographically large countries and in countries with substantial geographical variation in the supply of capital, as well as in economies where deposit-financed institutions play a large role in financial intermediation. Several extensions of these results would illuminate the effect of capital market segmentation. First, along the lines of Kashyap and Stein (2000) and Campello (2002), are banks in low deposit areas more exposed to monetary shocks? Second, after deregulation, are areas with many small, local banks more exposed to local deposit supply? Third, does the local deposit supply affect loan pricing and the capital structure for local firms?