ŠĻą”±į>ž’ Z\ž’’’WXYwż’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ģ„ĮE@ ųæėgbjbjƒęƒę +œįŒįŒhY%]’’’’’’ˆĢĢĢĢ   ­­­8>­d¢­lķŚ¶²d~µ.¬µ¬µ¬µ¬µŹ¶T·,ŲŲŲŲ/FŲZŁnŚ$£ŪRõŻ¬’Ś æø¬µ¬µæøæø’ŚĢ"ī¬µ¬µ§ŚćæćæćææøB ¬µ ¬µŲćææøŲćæ–ćæyĄ  yĄ¬µ² `}RžįXĀ­½@yĄ™Ą|½Ś0ķŚyĄ”ŽA½€”ŽyĄĢĢ  ”Ž yĄ J·>ˆ·,ćæ“·$Ų·ēJ·J·J·’Ś’Śd‡‚‹„!Įæ"‚‹ The Determinants of Bank Loan Pricing David O. Beim Columbia Business School 212-854-3484 March 20, 1996 A new dataset and a nonparametric methodology permit a detailed look at the many factors which affect the pricing of bank loans, clarifying the weight and significance of each. The data include the lending banks’ internal measure of each loan’s default risk, incorporating their private information about the borrower as well as their knowledge of each loan’s covenants and security structure. A variable for loan renewal and a study of repeat borrowing by the same companies offers new insight into the pricing impact of bank relationships. The most surprising variable is also the most significant: whether the loan is priced against Prime Rate or a market index such as Libor. Controlling for the difference in level among such benchmarks and for all the other explanatory variables, Prime-based borrowers pay about 140 basis points more. This remarkable anomaly persists even when a choice of benchmarks is offered to the same borrower. Introduction The pricing of corporate bonds, both in theoretical models and in Wall Street practice, is determined by a relatively small number of variables. The primary determinants are the term structure of riskfree rates and the default risk of the borrower. Models, following Merton (1974), most often posit the ratio of the market value of a company’s assets to its outstanding debt as the measure of default risk. Practitioners rely on the bond’s credit rating, which takes into account not only the volatility and leverage of the underlying business but also the seniority of the bond and any restrictive covenants it imposes on the company. If we are armed solely with a bond’s credit rating and a riskfree interest rate for its term or duration, we can come reasonably close to estimating the interest rate required on a risky bond. Bank loans are priced very differently, and respond to a much larger number of variables. Bank loans are generally priced at floating rates of interest, using a spread above a time-varying benchmark such as the Prime Rate or Libor. The spreads vary only moderately with credit rating and hardly at all with term. Equally or more important than risk and term are a bewildering variety of other variables. Previous studies of bank loan pricing have studied different sets of explanatory variables using OLS regression. In these studies, the object of interest was one or a few of the explanatory variables; the purpose was to explore a particular thesis about a particular variable rather than to expound bank loan pricing generally. The additional explanatory variables used in these studies were typically dependent on availability in the dataset used. Berger and Udell (1990) studied the Federal Reserve’s periodic surveys of terms of bank lending. They measured spread as the difference between the reported interest rate and a Treasury security of comparable duration. They found significance at the 5% level in the log of loan size, the log of duration, the presence of collateral, whether the rate is floating or fixed, whether the loan was made under a formal or informal commitment, whether the loan is a demand note, whether it is booked offshore, whether it is a participation, whether it is based on the Prime rate, the Federal Funds rate or a foreign money market rate (i.e. Libor), and whether its purpose is for various types of real estate construction. The R2 in this study for various dates ranged between 0.19 and 0.33. Booth (1992) separately regressed spreads on Prime-based, Libor-based and CD-based loans in a dataset composed from several sources and found significance at the 5% level in at least some cases in the following variables: log borrower sales, whether the loan is secured, the logarithm of the loan size, the amount of fees charged on undrawn balances, whether the loan arises from a loan commitment, whether the loan contains a benchmark option, whether the loan arose in a restructuring or an LBO, and whether the borrower is privately or publicly held. In a subset of public companies only, the following further variables showed significance: whether the firm had public debt outstanding, the logarithm of a numerical representation of the credit rating carried by the public debt, and whether the public debt is senior to the bank loan. His R2 values ranged from 0.23 to 0.42 in the first regressions and 0.25 to 0.47 in the second. Petersen and Rajan (1994) combined fixed and floating rate loans in 1988-89 survey data gathered by the SBA and the Federal Reserve. They found that these variables are significant for aggregate interest rates at the 5% level: the Prime Rate, the spread of BAA bonds above Treasuries, whether the loan is floating or fixed, log borrower assets, the rate of borrower sales growth, its coverage of interest by profits, the mean gross-profits-to-assets ratio for the industry, the age of the firm, and the number of banks from which the firm borrows. Their R2 values ranged from 0.145 to 0.158. Berger and Udell (1995) used the same survey dataset but selected for analysis the spreads over Prime-based lines of credit only. Of the 22 variables studied, 5% significance was found in only two: whether the borrower is a non-Subchapter S corporation and the log of the length of the relationship with the current lender. Their highest R2 was 0.095. This paper takes advantage of a large and comprehensive dataset of bank loans collected by Loan Pricing Corporation. Among the many explanatory variables in this dataset are the type of each loan (e.g. revolving credit, term loan, standby letter of credit, etc.), the purpose of each loan (e.g. for acquisition, recapitalization, working capital, etc.), the identity of the lending bank, the date the loan was signed, the presence of collateral, the magnitude of four classes of fees and the sales, industry, geographic location and the debt market access (bonds and/or commercial paper) of the borrowers. The dataset also contains a uniquely interesting variable: the banks’ internal measure of borrower risk. We believe that banks obtain private information about their borrowers, and such information must affect spreads. Such private opinion is likely to explain loan pricing better than could any “objective” standard of default risk. The banks’ risk measures also incorporate the effect of collateral and covenants in the loan agreements, i.e. like bond ratings they rate specific instruments rather than the borrower as a whole. A preliminary task in the present study to is map all the internal risk measures onto a common scale. This paper has three objectives. The first is to document, as comprehensively as possible, the relative importance of the many variables which affect bank loan pricing, and to discuss why the many variables should have the importance and the directionality that they do. I use a nonparametric procedure, alternating conditional expectations (ACE) as developed by Breiman and Friedman (1985). This procedure reveals a number of significant nonlinearities in the dependencies of spreads on continuous variables, and enables a parsimonious representation of the dependencies of spreads on discrete variables. The R2 of the regression of ACE transforms on spreads is 0.697. By itself, this might be merely an artifact of nonparametric estimation. Closer inspection shows that it results primarily from the quality of the data and not the methodology. An OLS regression of spreads on just four of the variables (reported in  REF _Ref351538870 \* MERGEFORMAT Table 6) has an R2 of 0.552. Local factors such as the borrower’s geography and industry and the identity of the lending bank turn out to have much greater importance than most economists would expect. Loans which are small relative to the scale of the company are priced at higher interest rates than those which are quite large. The study reveals a degree of inefficiency in the loan pricing market not previously documented. The second objective of the paper is to investigate the consequences of bank relationships for loan pricing. In recent years a great deal of theoretical and empirical work has been directed to the importance and consequences of bank relationships with borrowers. Papers such as Sharpe (1990), Rajan (1992), Petersen and Rajan (1993) and Boot and Thakor (1994) have modeled banking relationships in ways which make pricing inferences possible. Empirical work includes notably Petersen and Rajan (1994) and Berger and Udell (1995). Relationships are here studied in two ways. The first is through an indicator variable for loan renewal in the dataset. The second way involves reviewing all cases in which the same borrower is found taking more than one loan in the dataset. The pattern of spreads, depending on whether the borrower changes banks or stays with the same bank, offers new insight into the importance of relationship to bank pricing. The third objective of the paper is to examine in detail the most striking of all the explanatory variables: whether the loan is priced against the Prime Rate or a market index such as Libor. Taking into account the difference of levels between these benchmarks, and controlling for the many other variables which affect spreads, Prime-based borrowers in these data pay on average about 140 basis points (bp) more for their loans than Libor-based borrowers, with similar spreads to other market indices. The Prime pricing system and the market index system coexist in the same banks and the same cities with borrowers of similar size and risk. There is not a smooth transition from one system to the other, but a clear discontinuity. The discontinuity persists even when a choice of benchmarks is offered to the same borrower. These results beg explanation. After all, loans based on Prime and those based on a market index are very similar in character. The existence of substantially different prices for these nearly-identical products from the same vendors in the same markets is an economic anomaly of considerable interest. What can explain it? Several possibilities are explored: whether Prime borrowers have some unobserved difference from market index borrowers; whether Prime borrowers rationally pay more to avoid the greater volatility of market indices; and whether the Prime premium represents a subtle and systematic way of recovering informational costs and/or charging quasi-monopoly rents. The paper proceeds as follows. Section I describes the data and the variables available for study. Section II describes the methodology used and presents the basic results. Section III considers repeat borrowers and the relationship issue. Section IV studies in more depth the Prime premium and the possible explanations for it, and Section V concludes. I. Data and variables The data used in this paper come from the private loan database of Loan Pricing Corporation (LPC). They describe 75,162 corporate loan facilities signed between January 1, 1990 and December 31, 1995. Of these, 62,161 loans were based on a single benchmark and 13,001 offered the borrower a choice of benchmarks. Most of the analysis is performed on the first set of loans, although the second are analyzed separately in Section IV. Information on these loans was gathered by LPC primarily in the course of portfolio valuations performed at various times for 125 banks. In these valuations, banks pay LPC to evaluate a block of loans. Such a block will typically comprise all the bank’s loans which meet certain parameter tests, such as signed during 1995 and of more than $100,000 commitment size. The lending banks are typically part of large regional holding companies. Absent from the sample are a handful of the very largest bank holding companies and the many thousands of small banks which characterize the American banking system. Nevertheless, the data represent a cross-section of American bank lending which is believed to be broadly-based and reasonably representative. A. Lending benchmarks and spreads The most commonly used benchmark for floating-rate loans to major borrowers is Libor, the London Interbank Offered Rate for Eurodollar deposits. Libor is announced by the Bank of England each business day and represents a composite of the offered rate for Eurodollar deposits by major banks. Libor is the universal benchmark of international syndicated lending, and has become very popular in domestic U.S. loans as well. Libor is in fact more than one number: it is a set of rates, a small yield curve for deposits of 1, 3, 6 and 12 months term. Most bank loans in the U.S. market require quarterly payments of interest and so are priced against 3-month Libor. The second most commonly used market index benchmark is the U.S. domestic Certificate of Deposit (CD) rates. Cost of Funds (COF) measures are a variant on CD rates. Treasury Bills, Federal Funds and Bankers’ Acceptances are occasionally used but are less popular. All of the above indices are publicly available market prices. While a few banks use more exotic benchmarks, the present study is restricted to the market indices mentioned in this paragraph, plus Prime Rate. Prime Rate is sometimes described, by both academics and practitioners, as “the rate charged by banks to their most creditworthy borrowers.” Nothing could be further from the truth. As will be shown in Section IV, Prime has become the basis for most loans to the smallest and weakest borrowers. Large public companies have a strong tendency to borrow against market indices, and medium-sized companies increasingly do so as well. Prime Rate is unlike any other benchmark in that it is determined by the banks themselves. Furthermore, virtually all banks set exactly the same Prime Rate. When one bank changes its Prime Rate, either all other banks follow or, within a few days, the bank which changed reverts to the national standard. See Mester and Saunders (1995) for a discussion of how Prime Rate is changed. Figure 1 shows the movements of Prime, Libor and CD rates over 1974-1995, and Figure 2 shows the differences between these rates in the same time frame. The relationships among rates were somewhat disordered in the first half of the period shown, but in recent years clear patterns have emerged. The premium of Libor over domestic CD rates has settled down to about 1/8 of 1%, so that the two indices have nearly converged. Prime Rate, however, has drifted steadily further above Libor and CD rates, as emphasized by the regression line. The upward slope of this line is about 11 basis points per year.  REF _Ref325864433 \* MERGEFORMAT Table 1 gives descriptive statistics for the three rates and their monthly changes during this period. The market indices are somewhat more volatile than Prime, as can be seen from the standard deviations of monthly changes. The dependent variable in this study is the loan spread above the floating benchmark. To put all loans on a common basis, one must correct for the difference in level among the benchmarks. For example, if Prime stands at 2.50% above Libor, then a company which pays a spread of 0.50% above Prime might be expected to pay 3.00% above Libor. In order to make spreads comparable, all are adjusted to a Libor-equivalent basis by adding to them the average difference, in the month in which the loan was signed, between their pricing benchmark and Libor: Adjusted spread = Spread above Benchmark + (Benchmark-Libor)signing date In the above example, the spread above the Prime benchmark was 0.50%, Prime-Libor was 2.50% and so the adjusted spread was 3.00%. The adjustment to Libor-equivalent basis might be made differently. Since borrowers do not in general have the option of moving between Prime and Libor pricing, analysis should take into account not merely the Prime-Libor difference at the time the loan is signed, but the expected Prime-Libor difference over the life of the loan. Given the upward drift of the Prime-Libor difference, the expected future difference should be higher than the present difference. The ideal measure for this more sophisticated adjustment would be a Prime-Libor swap of a term matching that of the loans. Although such swaps exist, data on them are difficult to obtain, particularly with so many different terms and time points. Furthermore, this more sophisticated adjustment is less conservative in that a larger Prime-Libor adjustment would make Prime-based loans appear more expensive, thus making one of this paper’s central findings easier to prove. An important feature of Prime-based loan spreads is the tendency of unadjusted spreads to cluster at the point of zero spread, pricing sometimes referred to as “Prime flat,” so that the adjusted spread is simply the Prime-Libor difference. Of the pure-Prime based loans in the data studied, about 28% have zero unadjusted spread. The name “Prime Rate” and the fiction that this is the rate charged to the bank’s most creditworthy customers doubtless tend to enforce this barrier. However, about 2% of Prime-based loans have negative spreads, so that the barrier at Prime flat is not absolute. B. The borrower risk measure A unique feature of the LPC database is that it contains each bank’s internal scoring of each loan’s default risk. Every bank has some sort of linear risk rating system, typically using a numerical scale such as 1..6 or 1..10. Unfortunately, no two banks use the same system; even when the scale numbers are the same, the meaning assigned to the numbers may be quite different. The dataset does, however, record each bank’s rating of each of its loans. The problem of scale consistency among banks is handled in the following way. A standardized scale of 1..100 is created, with 1 corresponding to the lowest default risk. It is assumed that all banks observe this risk scale, and agree about the correct measure on this scale to be assigned to every loan. Where banks differ is in mapping different segments of the 1..100 scale onto their local scales. Each bank’s local risk scale is represented by a set of brackets on the 1..100 scale. A particular bank might assign 1(risk<23 to its category “1”, 23(risk<35 to “2” and so forth. Such a bank’s scale would be represented by the set of breakpoints between brackets { 23, 35, ... }. A complete collection of these sets of breakpoints, one for each bank, is termed a concordance. The concordance used in this study was constructed in the following manner. First, a “verbal concordance” was created as a first draft. This utilized the words used by banks in their own descriptions of their scales, matching them to a standardized verbal description for points on the 1..100 scale. This “verbal concordance” was discussed with most of the banks by LPC. If during the period studied a bank had changed its scale, or the definition used to describe its scale, it is treated as two separate banks, one before and one after the change, for this purpose. Refinement of this draft concordance exploited the fact that the dataset contains 20,371 pairs of loans to the same company by two different banks within the same 12-month period. (Note that if n banks participate in the same loan, n(n-1)/2 pairs are created). These paired ratings are used to improve the alignment of the separate bank scales against the 1..100 standard. It is assumed that the covenant and security structure of the paired loans is comparable, although this cannot be separately confirmed. The following loop was iterated: 1. Use the draft concordance to estimate the true risk of each loan in the paired-rating list. If the brackets assigned by the two banks overlap, use the midpoint of the overlap zone. If the two brackets are disjoint, use the midpoint of the space between them. If one bracket contains the other, use the midpoint of the inside bracket. 2. Use the loan risk numbers to re-estimate a draft concordance. This is done bank by bank with ordered probit analysis. The estimated risk numbers of all loans in the paired list rated by a particular bank are treated as the independent variable, and the bank’s local risk rating is treated as the dependent variable. Ordered probit provides a maximum likelihood estimate of the breakpoints in the independent variable. If ordered probit fails because the data are too sparse or disordered, breakpoints are estimated using mean risk values in each rating category. After five iterations the concordance changed little and the procedure was stopped. It is natural to compare the bank ratings with bond ratings in the 3,243 cases where the latter exist. This relationship turns out to be nonlinear and is graphed in Figure 3. Banks treat loans to companies with bond ratings of AAA through AA- as essentially equivalent. The curve then rises about linearly through the investment grade ratings, but flattens noticeably below BBB; indeed it declines for the lowest-rated junk bonds. This flattening and decline reflect the fact that in most cases banks require both collateral and strong covenants when they lend to the lowest-rated companies, giving them more safety than bonds or unsecured loans to better companies. It is clear from the pattern of Figure 3 that the risk ratings incorporate the effect of collateral and covenants. C. The borrower size measure The borrower’s size is an important determinant of spreads: small companies pay larger spreads than large companies, other things equal. Why should this be? At least three possible reasons come to mind. First, size may proxy for an additional dimension of default risk, since it is widely believed that small companies are less proven than large ones, more volatile and subject to sudden reverses. Second, smaller companies may be more expensive to monitor than large ones, per dollar of loans, since the amount of effort needed to monitor a company is unlikely to rise proportionately with the size of the company or of a particular loan. Third, small size may represent a lack of bargaining power on the borrower’s part, so that the excess spread charged to small companies is a form of rent unrelated to risk or monitoring cost. It is difficult to sort out which of these reasons is the correct explanation, or whether perhaps all three of them play a role. In this study the size of borrowers is measured by the logarithm (base 10) of borrower sales. Although total assets is an alternative measure of borrower size, almost all banks use sales as their size measure. Base 10 is used because many banks are organized into departments which handle companies of different sizes, often demarcated by a power of 10, and such institutional differences may have an impact on spreads. For example, the “middle market” of borrowers is often defined as companies with sales between $1 million and $100 million. It is convenient to see these demarcations as powers of 10. D. Loan type Loan type is a discrete variable classifying whether the loan is a term loan, a demand loan, a revolving credit, a standby letter of credit, an uncommitted guidance line, etc. The differences among these categories reflect different options on the part of the bank and the borrower as to when the loan is drawn down, prepaid, and/or redrawn. For example, a term loan most nearly resembles a bond, with a fixed drawdown and repayment date, although as in virtually all bank loans the borrower has an option to prepay at any time without penalty. In a demand loan, the bank has the option to demand repayment at any time. In a revolving credit, the borrower can increase and/or reduce balances outstanding at will over the life of the commitment. A standby letter of credit is not expected to be drawn down at all, except in emergencies. An uncommitted guidance line expresses an interest in lending, but not a legal obligation to do so. To derive the worth of these options one would need an underlying model of the borrowers’ cash flow, cash needs and leverage, which is well beyond the scope of this paper. Nevertheless, it is not at all surprising that loan type should have a bearing on the lending spread, i.e. that many of these options have measurable value. Understanding these option values is a fertile area for future research. E. Loan purpose Loan purpose is a discrete variable indicating whether the loan proceeds are used for working capital, acquisition, debt repayment. stock buyback, commercial paper backup, leveraged buyout, leveraged recapitalization, etc. In general, the difference among these purposes lies primarily in the different level of possible future default risk they imply. For example, a company entering into a leveraged recapitalization will become temporarily riskier, until its leverage returns to normal. On the other hand, a CP backup line does not increase the leverage of a company at all, since it would only be drawn down to repay CP. The purpose variable in effect creates an adjustment to the default risk premium to accommodate these differences. F. Industry Industry is measured in this study by the borrower’s 2-digit SIC code. Why should this be a determinant of bank lending spreads? The most plausible answer concerns differences in aggregate loan demand among various industries. For example, the cable television industry tends to borrow very heavily, and cable companies may have to pay a premium given the desire of banks to diversify their portfolios. Conversely, pharmaceutical companies rarely borrow, and probably obtain particularly low spreads given their potential to diversify loan portfolios. G. Geography One of the more striking characteristics of the bank loan market is its variation among cities, states and regions. Not only are spreads higher or lower in various areas, but the particular shape of the dependency of spreads on variables such as risk and size can be totally different from one city to another. In this study, geography is represented by two variables: the state and the Metropolitan Statistical Area (MSA) of the borrower, a discrete variable. It is quite unclear what economic characteristics of the states and MSAs account for these differences. A plausible candidate, for example, is the degree of bank market concentration, as measured perhaps by the Herfindahl-Hirschman index. Yet preliminary studies, not reported here, suggest that by itself this index has little explanatory power for the differences in loan spreads. This area requires substantial further research. H. Lending bank identity Another important explanatory variable, often overlooked, is the identity of the lending bank. It turns out that some banks price loans systematically above the average and some systematically below the average. This may reflect competitive issues such as the nature and quality of ongoing lending relationships or the entry of a bank into a new region. It may also reflect bank-specific characteristics such as different deposit rates, different operating costs or different portfolio strategies. This also requires further research. I. Signing Date Like all markets, the bank loan market shows changes in level over time. It has been widely reported that a “credit crunch” prevailed in 1990-1992, during which loan availability was reduced and bank risk aversion increased during a protracted recession. After that time, as recession gave way to economic recovery, banks again began to lend aggressively. Thus one would expect bank loan spreads to vary with time. The time variable in this study is the signing date of each loan. J. Term Term to maturity has an important influence on bond spreads, so it might be expected to similarly influence bank loan spreads. An important difference between the markets, however, is that bonds are generally fixed-rate instruments whereas bank loans are generally at floating rates of interest. A primary reason why bond prices are term-sensitive is the bondholder’s exposure to inflation risk. A bank is largely immunized to this risk by having loans with repricing maturities generally similar to those of its deposit and other liabilities. Even if interest rate risk is immunized, however, theoretical models of risky debt such as Merton (1974) and the many further papers based upon its general approach predict a relationship between spread and term; in particular, the spreads of very risky bonds are predicted to decline with term and those of very safe bonds are predicted to rise, with a humped pattern in the middle. These patterns were empirically confirmed in a general way by Sarig and Warga (1989). Thus, even in the bank loan market, one could expect the loan’s term to influence spreads. In this study the logarithm (base 10) of term expressed in months is used as an explanatory variable. K. Renewal The dataset contains an indicator variable which is one if the loan is a renewal of a previous loan agreement and zero otherwise. This variable offers a proxy measure of ongoing banking relationship. As discussed more fully in Section III, a number of theoretical models of bank relationships predict that over time banks may increase rates for existing customers, while other models predict that rates banks should lower rates as the relationship proceeds. The coefficient for loan renewal should provide insight into the this question. L. Loan scale Bank loan spreads are affected by how large a loan is relative to the size of the borrower, in other words whether the loan is a major or minor event in the company’s life. This variable is represented in this study by the logarithm (base 10) of the ratio of loan commitment size to company sales. A relatively large commitment might signal increased default risk, so one could expect a positive impact of loan scale on spreads. On the other hand, a company with a very large loan to negotiate may exert extra effort to bring in new lenders and increase competition, and banks might be eager to lead the company’s most important loans because of the future benefits that may accrue to the company’s most important banks. M. Capital market access Commercial paper (CP) is a close substitute for bank lending. Furthermore, CP trades at an interest rate quite close to the deposit rates of banks (Libor or CD rate). But because banks need to lend at a rate materially higher than their deposit rate, to create a profit margin, bank lending rates are necessarily higher than CP rates. This means that borrowers which have access to CP markets will generally utilize that access rather than borrow from any bank. When they do borrow from banks, their capacity to negotiate lower lending rates should be significant because of the viable CP alternative. Bond market access has similar properties, except that bonds are not necessarily cheaper than bank loans, when their fixed rates are swapped into floating rate equivalents. Nevertheless, companies with bond market access may also enjoy increased negotiating power when dealing with banks. In this study, CP and bond market access is proxied by two indicator variables which are one if the borrower has a CP or bond market rating and zero otherwise. The indicator variables are more useful for this purpose than the ratings themselves, for our object is to measure market access rather than default risk. N. Collateral The LPC database contains an indicator variable for whether the loan is collateralized or not. This is somewhat imperfect, since we do not know how much the collateral is worth compared to the loan amount nor how liquid the collateral it might be. Clearly a loan fully collateralized with cash would be essentially riskfree, while one collateralized with real estate might be quite troubled. Earlier studies such as Berger and Udell (1990) found that, contrary to intuition, the presence of collateral tends to be associated with higher spreads, not lower ones. O. Covenants All bond and bank loan agreements contain a number of covenants, some of them “boilerplate” and some crafted to the particular borrower. The purpose of such covenants is typically to lower the lender’s risk by making certain types of behavior by the borrower an act of default on the loans. A number of papers beginning with Smith and Warner (1979) have studied the possible or actual impact of covenants on lending spreads. Covenants may be conceptually grouped into those which are priced and those which are unequivocally required and therefore not priced. For example the negative pledge, which prevents the borrower from offering liens on assets to other lenders without ratably securing the loan in question, is so nearly universal that few unsecured borrowers would take the risk of being without it. On the other hand, a covenant requiring that debt not exceed a certain fraction of net tangible assets might be priced in the sense that if the borrower agreed to a lower fraction the lender might agree to a lower spread. Large syndicated bank loans frequently contain a sliding scale of spreads linked to either the borrower’s bond rating or alternatively to a ratio such as debt to cash flow. Pricing schedules such as these offer direct insight into the lenders’ demand function for the company’s debt if different leverage ratios prevail, and are worthy of separate study. The private LPC database, which is the subject of this paper, does not contain covenant information. However, as noted under Subsection B above, the risk-reducing effects of covenants are thought to be incorporated into the bank’s private risk assessment of the loan. In this way they enter into the study, even though their effect cannot be isolated. P. Fees It is often imagined that banks price loans on a “total return” basis, which takes into a account both the lending spread and the fees charged on the loan. If this were strictly true, higher fees would be associated with lower spreads, other things equal. Studies in the LPC database not reported here show that, on the contrary, higher fees in each of four categories are unequivocally associated with higher spreads. In other words, the conditions which make possible higher spreads also make possible higher fees. Because of this finding, fees are omitted as a variable explanatory of spreads: the causality appears to run from explanatory variables to both spreads and fees. Descriptive statistics for the continuous variables and bivariate discrete variables used in this study are displayed in  REF _Ref351786210 \* MERGEFORMAT Table 2. In addition to aggregate statistics, separate means for the subset of Prime-based loans and market-index loans are displayed. II. Methodology and Results Earlier studies of loan pricing have generally regressed loan spreads on explanatory variables using ordinary least squares. Two factors argue against the use of OLS in this study. First, as will be shown in this section, loan spreads may display important nonlinearities in their dependence on at least some of the five continuous variables studied here. Insofar as these nonlinearities are material, OLS is misspecified. Second, OLS requires a separate regressor (dummy variable) for each value of the discrete variables. We have 20 loan types, 21 purposes, 241 MSAs, 125 banks and 100 SICs, which would make a total of 507 new regressors. This creates not only an unwieldy computational problem but also a problem of perfect multicollinearity, as the dummies for each category sum to a vector of ones. While this can be alleviated by eliminating some of the dummies based on preliminary tests of significance, such a procedure is arbitrary at best, as most of the discrete values really do appear to matter. A preferable approach is to use a nonparametric model. I utilize a version of the Alternating Conditional Expectations (ACE) technique of Breiman and Friedman (1985). The ACE model is a generalization of the linear regression model y = ( + ((ixi+ ( into a form which still separates the variables but allows them a general functional form:  EMBED Equation.2  (1) where y is the dependent variable, the xi are the explanatory variables and ((y) and the (i(xi) are optimal transforms of their arguments, i.e. those transforms which maximize the correlation of the functions, and ( is the error term. Equation (1) is estimated using the algorithm described by Breiman and Friedman (1985). It requires alternating estimations of ( as the expected value of the right-hand side of (1) conditional on y and of each (i(xi) using the current estimates of the other functions to form conditional expected values of the function being estimated. To estimate the conditional expected values of continuous functions one must utilize a smoothing procedure. The procedure used in this study is linear kernel smoothing in which the window size is optimized using local cross-validation as described in Härdle (1990). The function ((y) is approximately linear when (1) is estimated with these data. In the results which follow, ((y) is assumed linear so that it can readily be inverted to recover y without losing the separation of the variables, and so that the (i(xi) can be read directly as basis points. The modified model then takes the following form:  EMBED Equation.2  (2) In this equation  EMBED Equation.2  is the unconditional mean spread. The (i(xi) are constrained to be mean-zero, which prevents a multicollinearity problem from arising. The sensitivity of y to each of the values of a discrete xi is expressed as a set of additions to or subtractions from the mean, one for each of the values, which sum over the dataset to zero. Dummy variables are not needed. The error terms ( are then mean zero by construction. The output of the estimation process is a set of relative-spread functions (i(xi). For continuous variables these are continuous functions and are displayed in Figures 4-8. For the discrete variables we obtain tables of relative spreads for each value of each discrete variable. To check the significance of the model and its components, we can use these functional outputs to map each xi onto its transform (i(xi), and then use OLS to regress y- EMBED Equation.2  on the transforms. The betas on the transforms all come out to approximately one but the R2 and t-statistics are of interest. We can assess the relative weight of each variable in determining spreads by observing the mean value of |(i(xi)| for each variable i over all observations. These results are displayed in  REF _Ref351533519 \* MERGEFORMAT Table 3. In this table, the first two columns indicate the largest effect the variable can have, while the latter two columns indicate the magnitude of the variable’s average or overall effect. The most important variable in terms of t-statistics is the first: the benchmark against which the loan is priced. Because of the importance of this variable, the detail of the ACE function for benchmark is displayed in  REF _Ref351533532 \* MERGEFORMAT Table 4. These numbers are relative spreads and are most meaningfully interpreted by taking the difference between any two. For example, Prime-based loans have adjusted spreads 17.5 bp above the mean of the dataset, but this has little meaning taken alone and simply reflects the high proportion of Prime-base loans in the dataset. The distance between this figure and any of the figures for market-index loans is more significant. For example, the table tells us that Prime loans have adjusted spreads 141 bp higher than Libor-based loans. Since Libor is the most common market index, this figure is the best single measure of the premium paid by Prime-based borrowers. There are significant differences among the market indices, for reasons that are not at all clear. It appears, for example, that those few borrowers who were able to negotiate pricing against Federal Funds or bankers acceptances benefited materially and those who borrowed at spreads above Treasury bills did less well. The reason for such differences is far from clear. But the most important gap is the one between any of these indices and the Prime Rate. This premium is explored in Section IV. After benchmark, the next most important variables in terms of mean absolute value and also t-statistics, are risk, size and the signing date. The risk curve in Figure 4 is monotonic and increasing up to about risk 60, which is approximately the limit for new market loans. The four risk deciles above 60 correspond to the regulatory categories of special mention, substandard, doubtful and loss, indicating a high proportion of troubled and renegotiated loans. The irregular pattern of spreads in this region suggests that banks sometimes attempt to collect more interest in a renegotiated loan and sometimes are forced to give interest rate concessions. The downturn of spreads in risk 75-85 suggests such concessions. Note that the range of spread impact of risk over the market range of 1-60 is only about 80 bp, representing substantially less default risk discrimination than one sees in corporate bonds. Figure 5 shows the curve against the logarithm of sales, which is nearly linear. Note that the range of impact, about 140 bp between minimum and maximum, is greater for borrower size than for borrower risk, although the mean impact is about the same. Figure 6 shows the ACE transform for signing date. This curve displays the sort of random-walk pattern seen in many other market time series. Note that lending spreads rose materially in the first 18 months of the period studied, from early 1990 to mid 1991. This was the period of the “credit crunch”, in which banks made fewer new commitments due to a recession and major loan losses. From mid-1991 through 1995, loan spreads appear to have stayed generally high. Figure 7 shows the dependence of spreads on the loan’s term. Note that the effect is extremely small, only a few basis points, and that the shape is not particularly significant in that context. Note too that all the impact on spread occurs for terms of less than 24 months; beyond that point, term has no effect on spread whatever. It is likely that the result shown here is a blend of differing shapes for different loan types and risk groups. Further study is needed, but the overall conclusion is that term scarcely affects loan pricing. Figure 8 shows that loan scale, the logarithm of loan size / borrower sales, has a material negative effect: relatively larger loans are priced at lower spreads, contrary to what might have been expected. It appears that the risk-increasing effect of large loans is already captured in the risk measure and to some extent in the purpose variable. What is apparently shown here is the tendency of banks to compete harder to larger transactions, and for borrowers to exert more effort to stimulate such competition when they have a large amount to borrow, relative to their size. The most important discrete variable is the identity of the lending bank. It may come as a surprise to see such a large bank-specific effect, but the difference between the prices of the most aggressive and the most conservative lenders during this period is approximately 200 bp. Furthermore, the dispersion of bank policies is sufficiently great that the mean impact of this variable is also quite high. The largest gap between minimum and maximum response is the ACE function for MSA. Depending on its city, a borrower may experience different loan pricing over a range of more than 300 basis points. The more moderate mean absolute value, however, indicates that these extremes are confined to a relatively few cities. Why certain cities differ so greatly from the norm deserves a great deal of further study. To a lesser but still important extent, state and industry show similar patterns. Some states and some industries feature major departures from lending norms, but the mean absolute effect is moderate. All of these variables are weightier than loan type and loan purpose, which display similar patterns: a few types and a few purposes count heavily, while many matter much less. Consistent with Berger and Udell (1990), this study finds that collateral is associated with higher spreads. This is widely understood as reflecting a risk factor that may not be captured in the more conventional measures of default risk -- banks insist on collateral from companies they know to be risky. In the context of the bank-focused risk measures used in this study, it suggests that banks may reduce their internal risk assessment more than their spread when collateral is added to a loan. Finally, capital market access turns out to have hardly any effect at all on bank loan pricing, contrary to the view that such access ought to significantly affect banks’ willingness to offer interest rate concessions. Bond market access is literally without impact. Commercial paper access reduces loan rates enough to show statistical significance, but the impact is only 2.5 bp, which has little practical significance. III. Are bank relationships priced? The modern understanding of banking is based upon the economics of information. In this paradigm, banks gain private information about their borrowers and engage in costly monitoring of the borrowers’ activities. The impact on loan pricing arises in several ways: through the cost of information gathering and monitoring; through the improvement of incentives that may arise from monitoring, through the rents that may be associated with a bank’s quasi-monopoly on borrower information, and through concern for reputation by both banks and borrowers. Which of these is most important, and the directionality of the net effect, are not at all clear. Diamond (1984) pioneered the analysis of banks as delegated monitors of borrowers. He showed that monitoring is valuable in resolving incentive problems and showed how cost considerations lead investors to delegate this function to intermediaries with diversified portfolios. In Diamond (1991), firms build their reputations by agreeing to pay the extra cost of bank-monitored debt; firms which acquire good reputations then move on to the arms-length debt markets to save on monitoring cost. Petersen and Rajan (1993) developed a model in which banks offer higher rates to customers at an early stage of the relationship, when borrower types are unknown, then lower them later as information is revealed. Similarly in Boot and Thakor (1994), borrowers pay high rates and offer collateral in the early stages of the banking relationship, but are more favorably treated as their projects succeed. Both of these models suggest that rates should fall in the course of a banking relationship. In Sharpe (1990), banks gain information about their borrowers by the act of lending, which then allows them to capture rents from older customers. Competition leads new banks to offer initially lower rates. A bank’s concern for its own reputation moderates its exploitation of clients it has “informationally captured”. In Rajan (1992) banks are contrasted with arms-length lenders. An entrepreneur obtains improved incentives from bank borrowing with monitoring, but also pays a cost since the bank will extract rents over time as the relationship proceeds. An arms-length debt contract avoids the costs but also the benefits. Both of these models suggest that rates should rise in the course of a banking relationship. Of the empirical papers, Petersen and Rajan (1994) found that relationship length affected the available quantity of bank credit but not its price. In contrast, Berger and Udell (1995) found evidence that interest rates fall as a banking relationship proceeds. Predictions that rates rise or fall in the course of banking relationships can be tested in two ways with the LPC data. First, the data contain an indicator variable for loan renewal, which should be a proxy for ongoing banking relationship. The impact of this variable can be seen in  REF _Ref351533519 \* MERGEFORMAT Table 3. As in the case of commercial paper access, a statistically significant effect is found, but its magnitude is so small (2.8 bp) as to have no practical importance. Additional insight into the importance of relationships can be obtained by selecting all cases where the same company borrows more than once and separating those which borrow only from one bank from those with multiple banks. Of course, this approach is imperfect because the dataset does not contain all loans. Nevertheless, the dataset is sufficiently broad that the results are suggestive. There are 19,457 cases of borrowers with more than one loan in the dataset. These are divided initially into two groups: those in which all such loans are made by the same bank, and those in which two or more banks are involved. The first set is then grouped by risk rating, to control of any change in risk rating between lending dates. Each such subset is then further separated by benchmark, so that only loans made on the same benchmark are compared. Each such further subset is then examined to see whether unadjusted spreads rose or fell. This is determined by regressing spreads on loan signing dates, checking for positive or negative slope. The second set, with multiple banks, is then analyzed similarly except that no effort is made to ensure consistency of risk ratings among the various banks. The results are reported in  REF _Ref351537799 \* MERGEFORMAT Table 5 together with t-tests for significance of a difference in means between the two sets and the combined data. Borrowers with only one bank are more than twice as likely to experience no change in spreads as borrowers with two or more banks. The difference is statistically very significant. Conditional on spreads having changed, a decline in spreads is more likely than a rise regardless of the number of banks. But the proportion of changes which are rises in spread is not different between the two sets of cases. We cannot reject a null hypothesis that, conditional on spreads changing, the probability of spreads rising is the same for borrowers with one bank and those with two or more banks. In other words, borrowers with one bank are associated with stasis in spreads: the status quo is likely to be maintained. Situations of changing spreads are more associated with multiple banks, but such change may be either positive or negative. The frequency of positive changes, given that change occurs, is the same for one-bank borrowers and multiple-bank borrowers. This implies that having multiple banks is not distinctively beneficial nor harmful, because there are different reasons why a borrower may have multiple banks. In the “good scenario”, a company’s success and growth is noticed by new banks who then compete for its loan business, offering lower spreads as inducement. In the “bad scenario”, a company’s problems and failures make bank loans difficult to obtain from its traditional bank or banks, and it seeks new banks, offering higher spreads as inducement.  REF _Ref351537799 \* MERGEFORMAT Table 5 suggests that the good scenario is somewhat more frequent than the bad scenario in the sample being studied, but that the frequency with which either scenario occurs depends solely on events within the company and is not affected by the number of banks from which it currently borrows. This result is fully consistent with Petersen and Rajan (1994), who found that lending cost rises and credit availability falls with the number of banks in a sample dominated by very small firms. We need only assume that the “bad scenario” may be far more common than the “good scenario” in such a sample. One would expect the opposite result in a sample of large, public firms, given that significant success and growth is required to be in such a sample. The results offer little support for models such as Rajan (1992) and Sharpe (1990) which predict that spreads will rise with repeat borrowing. The coefficient on the renewal variable is positive and carries statistical significance, but its magnitude is insignificant. The repeat-borrower study shows why: repeat borrowers from the same bank tend strongly to maintain the same spread they obtained in the last borrowing. There is an important sense, however, in which banks do charge a rising loan rate to companies with which they have relationships based on private information. We need only suppose that relationship banking is more associated with the Prime Rate and that non-relational efficient-market banking is more associated with Libor and other market indices. This is certainly plausible as a first approximation, as discussed more fully in Subsection IV (C). The upward drift of the Prime Rate above Libor has the exactly the same economic effect as an increase in spreads. The adjusted Libor-equivalent spreads of Prime-based loans all tend to drift upward over time because the Prime-Libor difference drifts upward. It seems that the Prime Rate system may provide the predicted increase in a subtle and systematic way. This brings us to the puzzle of the Prime premium. IV. The Prime premium The ACE estimation found Prime Rate borrowing to be more expensive than Libor borrowing by 141 basis points in the data studied. To check the robustness of this result, two further exercises are performed. The first robustness check is an OLS regression of adjusted spreads on a constant, borrower risk and size, the signing date and a dummy variable which is one if the loan is Prime-based and zero otherwise. The results are displayed in  REF _Ref351538870 \* MERGEFORMAT Table 6. The parameter on the Prime dummy is 144 basis points, in close agreement with the ACE estimate, and its t-statistic is 162.0. The strength of the result seems to owe more to the quality of the data than to any particular econometric procedure. The second robustness check is a display of mean adjusted spreads in cells segmented by borrower risk and size, with a t-test for difference of means in each cell. This is show in  REF _Ref325864516 \* MERGEFORMAT Table 7. The t-tests are all highly significant, and the magnitude of the Prime premium can be seen as a joint function of risk and size. The magnitude is higher than 140 in some cells and lower in others, but the order of magnitude is not different. One might expect the Prime premium to decline in the larger size and lower risk categories. After all, this anomaly appears to be a kind of market inefficiency, and one would expect to find a more-nearly efficient market as one moves from the smallest and riskiest companies toward the largest and safest ones. Contrary to this expectation, the premium displayed is largest for the largest companies. On closer examination, one can see the reason for this pattern. Market index spreads do indeed fall as company size increases and risk falls, but adjusted Prime spreads tend to plateau at 260-280 basis points. This is just above the Prime-Libor difference, so we are seeing a preponderance of Prime-flat pricing. The adjusted Prime spreads do not fall because the banks’ resist charging negative (unadjusted) spreads on Prime-based loans. The resistance barrier at Prime flat helps make the Prime premium a step function: Prime-based and market index pricing do not flow smoothly into each other. The principal that Prime flat is the lowest Prime-based loan rate seems to dominate any need to make Prime-based loans competitive to the largest and strongest companies. An economic anomaly as large as the Prime premium demands explanation. This section discusses three possible explanations for it. The first is that borrowers in the Prime system may have certain unobserved characteristics which make them different from similar-seeming borrowers in the market index system. For example, they may be riskier in some way not captured by the risk and size measures, or they may require a degree of monitoring which is greater than other companies and not easily measured by outsiders. This explanation is explored in Subsection A. The second possible explanation, explored in Subsection B, is that Prime based borrowers rationally pay the Prime premium to avoid the greater volatility of market indices. Both of these explanations seem unsatisfactory on closer examination. A third explanation, based on a mix of generalized cost recovery and quasi-monopoly rents, is explored in Subsection C. A. Does the Prime premium reflect unobserved borrower differences? This subsection considers whether Prime borrowers might have certain unobserved characteristics, such as greater risk or monitoring cost, which put them in a different pricing category from market index borrowers. Certainly Prime-based borrowers are on average different from market-index borrowers. For example, the last two columns of  REF _Ref351786210 \* MERGEFORMAT Table 2 show that Prime-based loans are on average riskier (36 vs. 28) and are made to smaller firms (less than $1 million of sales vs. more than $100 million). About 28% of market index borrowers have bond market access and 20% have commercial paper access, versus 2% and 1% in the Prime population. Despite these differences on average, Prime-based and market index loans overlap in the “middle market” of companies with sales of $1-100 million.  REF _Ref351436565 \* MERGEFORMAT Table 8 sets out a distribution of the Prime-based and market-index loans by borrower risk and size. The bias of Prime loans to smaller and weaker borrowers is evident, but so also is the overlap of the two pricing systems in the middle market. For example, firms with sales between $10 million and $100 million reppresent about 39% of the Prime loan market but also 24% of the index loan market. The Prime premium is a paradox in this zone of overlap, where companies of similar size and risk obtain loans whose cost varies materially depending on which benchmark is used. The question is whether, in this overlap zone, there exist unobserved differences between Prime borrowers and index borrowers which might explain the Prime premium. It is not usually possible to rule out factors which by definition are not measurable. However, there is a subset of the data which was excluded from the results reported thus far and which casts considerable light on this issue. Banks often give the borrower an option to choose between a Prime-based interest rate and a market index-based interest rate. Because the borrower is the same, we can rule out unobserved differences among borrowers in these cases. Here, of all places, one would expect benchmark irrelevance to hold: if virtually identical products are offered by the same vendor to the very same buyer, how could there possibly be a pricing difference? Yet, remarkably, one can easily reject a null hypothesis of benchmark irrelevance even in the case of optional-benchmark loans, as is clear from  REF _Ref326639913 \* MERGEFORMAT Table 9. When Prime-based loans and market index-based loans are simultaneously offered to the same borrower, the Prime-based loans are on average 146 bp more expensive. This is the same magnitude observed in different data for unrelated borrowers. In only 273 out of 13,001 optional-benchmark cases is the adjusted cost of the Prime-based loan below that of the market index equivalent. In most cases, the pricing anomaly is so great that the bank could not seriously intend that the borrower choose the Prime pricing. The option has little or no value, particularly in view of the tendency of Prime to drift further above Libor and CD rates with the passage of time. Why should banks expend effort to offer an apparently valueless option? The behavior is rational only if banks need to preserve and promote the Prime pricing system as a system, regardless of its utility in particular cases. This “system” view of Prime will be pursued further in Subsection C. If the Prime premium were caused by unobserved differences among apparently similar borrowers, it would vanish in simultaneous offers to the same borrower. The fact that it does not vanish suggests that something else is causing the anomaly. B. Does the Prime premium pay for risk differences in the benchmarks? We must also consider the possibility that Prime-based loans are more attractive to borrowers. Can the Prime premium be explained by differences in the benchmarks themselves? As is clear from  REF _Ref325864433 \* MERGEFORMAT Table 1, market indices are somewhat more volatile than Prime: market indices change daily while Prime changes only a few times during the year. Is this sufficient to explain the spread differences? What should a borrower be willing to pay to avoid increased volatility of interest payments? We can estimate an answer to this question because companies frequently pay extra interest to borrow at fixed rates for medium or long terms in the bond market. The term premium may be viewed as a measure of risk aversion by bond issuers. What matters is not the raw volatility of rates but the interaction between this volatility and a borrower’s sampling frequency. A company which issues a bond once every five years samples bond rates at five-year intervals. A company which borrows against three-month Libor samples Libor once every three months. In short, we cannot separate the volatility question from the variability of actual interest payments under some borrowing strategy. There is an unlimited number of possible borrowing strategies, so it is difficult to prove a general result. One can, however, simulate several typical borrowing strategies and examine the results for order-of-magnitude differences. What follows, therefore, is suggestive rather than rigorous. The model borrower in this example carries a Baa/BBB bond rating. The great majority of bank borrowers do not have bond ratings, but the data studied in this paper contain 267 Prime-based loans and 379 Libor-based loans to borrowers rated Baa or BBB. If we restrict ourselves to the most numerous size class, those with sales between $100 million and $1 billion, we find 141 Prime-based loans whose average spread (unadjusted) is 20 bp and 197 Libor-based loans whose average spread is 78 bp. Three alternative strategies are simulated here over the 20-year period 1975-1994. The first is to maintain a bank loan at Prime+0.20%. The second is to maintain a bank loan at Libor+0.78%. The third is to sell four successive five-year notes at Moody’s Baa index rate. All interest payments are assumed to be quarterly; the bond index is adjusted for this timing difference from normal bond practice. A Prime loan with quarterly interest payments changes in cost on each day Prime is changed: borrowers pay interest in arrears based on the average Prime during the preceding quarter. A loan priced against a market index such as Libor, however, locks in a 3-month rate (plus spread) for the coming quarter at each payment date: borrowers pay interest in arrears based on the value the index carried three months before. Thus cash flows in this simulation use the three-month average Prime but the first-month Libor reading in each quarter from the Citibase data. The outcomes are summarized in  REF _Ref326641144 \* MERGEFORMAT Table 10, which shows the descriptive statistics for the three streams of interest payments and also for the difference between the Libor-based stream and each of the other two, labeled “Prime Increment” and “Bond Increment” respectively. Analytically, we start from the Libor-based stream, which has the lowest mean cost but the highest variance, and we consider two increments which raise the cost but lower the variance of interest. We know that many firms choose to borrow in bond markets precisely because they wish to make this tradeoff, yet many choose not to do so. We may reasonably assume, therefore, that the expected utility of the representative firm is not changed by adding or subtracting the Bond Increment. We can use this intuition to calibrate a simple model of risk aversion and ask what the same firm would pay for the risk reduction offered by the Prime Increment. Assume that an agent with exponential utility of coefficient ( has random one-period future wealth W and is indifferent to an additional, stochastically independent risk ( ~ N((,(2). Equalizing expected utility, we have  EMBED Equation.2  Since W and ( are stochastically independent, the left-hand expectation is separable, and we are left with  EMBED Equation.2 , that is, ( = ((2 / 2 (3) The Bond Increment has ( = 3.84 and ( = 12.62, so ( = 0.05. The Prime Increment has ( = 2.05; using the same ( = 0.05 we conclude that indifference would require ( = 0.10. In other words, a borrower with exponential utility willing to pay 3.84 to achieve the risk reduction of bonds should pay only 0.10 for the protection of the Prime Rate. The apparent payment of 2.05 seems about twenty times too high. To put it differently, the fact that Libor is more volatile than Prime seems in this example to account for only about 5% of the Prime premium. Another element of this simulation deserves mention. If the Prime premium pays for lower volatility, then it should have been most valuable in the period 1980-84, when Libor was particularly volatile and the Prime Increment’s standard deviation was highest. On the contrary, the Prime premium (in this case, mean of the Prime Increment) was low in this period and rose later. It seems we must look elsewhere for explanations. C. The dual bank pricing system The third explanation of the Prime premium views Prime-based lending and market-index lending as two quite separate systems, originally intended for quite different clienteles and informational environments, but now increasingly commingled. In this view, Prime-based lending is a system designed for information asymmetries, where banks engage in costly investigation and monitoring of borrowers. It is, in short, an institutionalization of the modern theoretical paradigm of banking. Economists have usually assumed that banks recover their informational costs and relationship investments borrower by borrower, charging each one a loan spread which somehow incorporates the cost of doing business with that company. What this overlooks is the institutional difficulty of truly understanding costs on a customer-by-customer basis. Very few banks claim to have this understanding. All they can readily see is the aggregate overhead. Thus a third possible explanation of the Prime premium is that it represents a generalized, institutionalized way for banks to recover their aggregate relationship costs from a class of companies which, broadly speaking, ought to be paying for these costs. A related explanation is that the premium represents rents extracted from a class of companies too weak to resist paying. It is difficult to separate rents from cost recovery in an environment where cost is not well-measured on a borrower-by-borrower basis. Prior to the emergence of modern, information-based theories of banking, economists tended to view banking as one component of an efficient financial market. For example, Black (1975) viewed the banking market as efficient except for the presence of regulation. He visualized unregulated banking as a few large, national banks operating at low unit cost against market indices such as the Federal Funds rate. He concluded that most bank policies optimal in an unregulated market are also optimal in the actual market because so many regulations can be avoided. The portion of the corporate world to which efficient-market banking would most nearly apply is the high end -- large, public companies with well-disseminated financial information, widely followed by stock analysts and the financial press. With few informational asymmetries to overcome, such companies would not require costly investigation and monitoring by banks, nor would they pay for it. Starting in the 1970’s, the largest banks began offering Libor-based loans to companies whose size and well-understood information made monitoring and relationship investment unnecessary. Such companies typically had the full range of securities alternatives -- commercial paper, bonds and stocks -- and were unwilling to pay banks much spread above market rates. The loans were often syndicated to banks which had no relationship with the borrowers. Virtually all syndicated loans are priced against Libor. The Libor system was, at least initially, an institutionalization of non-relational “efficient market” banking with such borrowers. At first, this represented an appropriate dichotomy of the market: Prime was paid by those companies with higher informational costs. As time has gone on, however, offering Libor-based loans to middle-market borrowers has become a competitive tool, a way for new banks entering a market to gain business and pressure competitors. The breakdown of geographic restrictions in U.S. banking during the 1980’s accelerated this aggressive use of Libor. Figure 9 illustrates the proportion of Prime-based loans in the dataset. To be sure, this is only a small fraction of the total bank loans in the United States. But in this sample at least, the decline in the Prime Rate system during 1990-1995 is dramatic. By number of loans, the Prime proportion declined from just over 90% to just under 70%. By dollar volume, the Prime proportion declined from about 70% to about 25%. This in turn explains the trend of Prime to ever-higher levels above Libor illustrated in Figure 1. As Libor usage expands, the remaining Prime-based borrowers become ever smaller and riskier. But this is a penalty to all previous Prime-based borrowers, and it must strengthen the resolve of those who notice it to break out of the Prime Rate system if they can do so. As the gap becomes wider, the anomaly becomes more glaring. It is not clear how much longer the Prime system can survive. We can apply the classic Stiglitz and Weiss (1981) analysis of adverse selection. Assuming that banks cannot clearly distinguish between the best and the worst of the Prime-based borrowers, the escalating rate will drive safer borrowers out of the system faster than it drives out riskier ones. This is turn can lead to an internal equilibrium interest rate: beyond a certain point, allowing the Prime Rate to drift higher may lower banks’ expected returns. This suggests that the Prime Rate system may be abandoned by banks at some point, as the risks of the remaining borrowers rise to the point that credit rationing begins. The modern theoretical paradigm of banking is quite new, having been developed almost entirely in the 1980’s and 1990s. It has quite fully replaced the earlier efficient-market view of banking. It is ironic that during the same period in which “informational” banking was replacing “efficient market” banking in the world of theory, exactly the reverse was beginning to happen in the world of practice. V. Conclusion This paper studies a new database with a new methodology to better understand bank loan pricing. It has three objectives. The first objective is to give the most comprehensive account yet offered of the many factors which influence bank lending spreads, showing the relative weight and significance of each. Factors which are important in bond pricing, such as borrower risk and loan term, have only moderate importance in bank lending, while other factors such as borrower size and geographic location, lender identity, and pricing benchmark have unexpectedly high significance. This paper documents a degree of inefficiency in bank loan pricing much greater than previously established. It is quite unclear why banks should differ so materially in their average spread levels. It is equally unclear why various MSAs should display such very different pricing patterns. Future research will be required to explore these anomalies. The paper’s second objective is to explore the impact of bank relationships on loan pricing, and in particular to test the hypothesis that banks which informationally monopolize borrowers tend to increase their spreads over time, as predicted by several theoretical models. The results suggest that loan spreads to borrowers with only one lender in the dataset tend to stay flat, not to rise. However, insofar as informational monopoly is associated with Prime-based lending, the spread increase does occur in a generalized way through the upward drift of Prime over Libor. The third objective is to explore more deeply the single most surprising empirical finding: that controlling for all the variables which seem to affect bank loan spreads, Prime-based borrowers appear to pay a great deal more for their loans. This was demonstrated in several different ways. Arguments that the Prime premium may reflect unobserved borrower differences or excessive volatility of market indices were considered but found unconvincing. An alternative explanation was offered: the Prime Rate system used to represent a generalized way in which banks recovered their cost of obtaining private information, building relationships and monitoring borrowers. Economists have long posited that banks recover these costs explicitly in their lending rates to each borrower; in fact banks appear to have done so subtly and almost invisibly through a Prime Rate which steadily rises above market rates. In recent years, however, Libor-based loans have been widely used as a competitive tool by banks entering new markets as geographic barriers break down. The Prime premium offers a tempting competitive opportunity. It seems only a matter of time until market indices complete their domination of the commercial loan market. If this occurs, the excess cost of Prime will continue to grow ever larger and be charged to an ever smaller circle of borrowers. A Stiglitz-Weiss end is foreseeable. References Berger, Allen N. and Gregory F. Udell, 1990, “Collateral, Loan Quality and Bank Risk”, Journal of Monetary Economics, 25:21-42. Berger, Allen N. and Gregory F. Udell, 1995, “Relationship Lending and Lines of Credit in Small Firm Finance”, Journal of Business, 68:351-381. Black, Fischer, 1975, “Bank Funds Management in an Efficient Market”, Journal of Financial Economics, 2:323-339. Booth, James R., 1992, “Contract Costs, Bank Loans, and the Cross-Monitoring Hypothesis”, Journal of Financial Economics, 31:25-41. Boot, A. W. A. and A. V. Thakor, 1994, “Moral Hazard and Secured Lending in an Infinitely Repeated Credit Market Game”, International Economic Review, 35:899-920. Breiman, Leo and Jerome H. Friedman, 1985, “Estimating Optimal Transformations for Multiple Regression and Correlation”, Journal of the American Statistical Association, 80:580-619. Diamond, Douglas W., 1984, “Financial Intermediation and Delegated Monitoring”, Review of Economic Studies 51:393-414. Diamond, Douglas W., 1991, “Monitoring and Reputation: The Choice Between Bank Loans and Directly Placed Debt”, Journal of Political Economy, 99:689-721. Härdle, Wolfgang, 1990, Applied Nonparametric Regression (Cambridge University Press). Merton, Robert M., 1974, “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates”, Journal of Finance, 29,449:470. Mester, Loretta J. and Anthony Saunders, 1995, “When Does the Prime Rate Change?”, Journal of Banking and Finance, 19:743:764. Petersen, Mitchell A. and Raghuram G. Rajan, 1993, “The Effect of Credit Market Competition on Firm-Creditor Relationships”, Working paper, University of Chicago. Petersen, Mitchell A. and Raghuram G. Rajan, 1994, “The Benefits of Lending Relationships: Evidence from Small Business Data”, Journal of Finance, 49:3-35. Rajan, Raghuram G., 1992, “Insiders and Outsiders: The Choice between Informed and Arms-Length Debt”, Journal of Finance, 47:1367-1400. Sarig, Oded and Arthur Warga, 1989, “Some Empirical Estimates of the Risk Structure of Interest Rates”, Journal of Finance, 44:1351-1360. Sharpe, Steven A., 1990, “Asymmetric Information, Bank Lending, and Implicit Contracts: A Stylized Model of Customer Relationships”, Journal of Finance, 45:1069-1087. Smith, Clifford W. and Jerold B. Warner, 1979, “On Financial Contracting: An Analysis of Bond Covenants”, Journal of Financial Economics, 15:175-219. Stiglitz, Joseph E. and A. Weiss, 1981, “Credit Rationing in Markets with Imperfect Information”, American Economic Review, 71:393-410. Figure 1 This figure graphs the monthly average values of Prime Rate, 3-month Libor and 3-month U.S. domestic Certificate of Deposit (CD) rates during 1974-1995 as reported by Citibase. Figure 2 This figure draws from the same data as Figure 2. It displays the difference between Prime Rate and Libor, and the difference between Libor and CD rate. The straight line is an OLS regression line of the Prime-Libor differences on time. The upward slope of this line is 11 bp per year. Figure 3 The 3,243 loans in the dataset with public bond ratings are divided by rating, and the bank ratings assigned to each such subset are averaged. The result is displayed as a graph of bank rating against bond rating. Figures 4-8 These figures graph the functions (i(xi) of Equation (2) estimated for each of the five continuous variables used in the study. The unit of measure is basis points -- the number of bp added to or subtracted from the mean adjusted spread by the variable in question. All the functions are mean zero over the dataset by construction. Since the functions are separable, each graph shows the dependence of adjusted spread on the variable in question, controlling for all other variables. Figure 9 The single-benchmark loans in the dataset are divided into monthly subsets according to their signing date, and further separated into Prime-based and market-index loans. The chart shows the proportion in each month which are Prime-based, both by number of loans and by dollar value of loans. Trendlines are generated byOLS regression on the month of signing. Table  SEQ Table \* ARABIC 1 Descriptive Statistics for Monthly Average Benchmark Rates, 1974-1995 This table shows the means, standard deviations and correlations with Prime of monthly average Prime Rate, 3-month Libor and 3-month CD rates as reported by Citibase. The statistics are also shown for the monthly changes in these rates.  MeanStandard DeviationCorrelation with PrimeInterest Rates Prime Rate10.0193.3881.000 Libor - 3 month8.6423.6160.967 CD - 3 month8.1633.3730.970Monthly Changes in Rates Change in Prime-0.0050.6691.000 Change in Libor-0.0120.8410.782 Change in CD-0.0110.8160.766 Table  SEQ Table \* ARABIC 2 Descriptive Statistics of Variables This table reports descriptive statistics for the continuous variables and binary discrete variables used in this study over the set of 62,161 loans priced over a single benchmark, together with means over the subsets of Prime-based loans and market-index loans. Prime-based loans represent 50,897 or 82% of the number of single-benchmark loans studied, but only 45% of the dollar volume of loans. All logarithms are to base 10.  -----------All Single-Benchmark Loans------------Prime-based LoansMarket-index Loans Variable  Minimum Maximum MeanStandard Deviation Mean MeanLibor-equivalent spreads -88.401285.22282.77107.60315.53134.72Continuous variablesBank Risk Measure0.00100.0035.1512.6436.3128.07Log Borrower Sales (mm)-3.004.001.190.970.952.27Signing Date (month)0.0071.9638.7017.6537.4844.22Log Loan Term (months)0.002.561.280.411.231.50Log (Loan Size / Sales) -5.703.21-1.130.72-1.04-1.55Binary Discrete VariablesRenewal0.001.000.480.500.490.39Collateral0.001.0047.4921.9047.5747.16Bond Rating0.001.000.070.250.020.28CP Rating0.001.000.050.210.010.20 Table  SEQ Table \* ARABIC 3 Results of ACE Estimation and Regression This table displays the minimum value, maximum value and mean absolute value of the functions (i(xi) in Equation (2) of the text. These functions are used to transform the data, and the transforms are regressed on adjusted spreads. The betas in this regression are all approximately one, as would be expected. The t-statistics are shown in the right column. The R2 of this regression is 0.697. ----------------ACE Coefficients (i(xi)----------- Continuous variables Minimum MaximumMean Absolute ValueRegression t-statisticBank Risk Measure-38.6369.6316.0752.16Log Borrower Sales (mm)-74.7262.5518.9052.68Signing Date-101.4438.3915.1965.36Log Loan Term (months)-3.995.153.306.59Log (Loan Size / Sales) -45.3528.434.9328.82Binary Discrete Variables Renewal-1.291.501.384.39Collateral-83.597.502.7039.90Bond Rating-4.220.150.310.23CP Rating -2.530.060.133.40Multi-valued Discrete Variables Benchmark-163.2918.5231.87120.32Loan Type-83.5957.852.7011.60Loan Purpose-82.3019.481.987.95Lending Bank-60.36125.2813.6446.34State-72.2151.135.6221.47MSA-144.74171.345.3825.88Industry-29.5742.506.2221.76 Table  SEQ Table \* ARABIC 4 Detail of ACE Function for Benchmark This table, an example of ACE output for a multivalued discrete variable, sets forth the ACE transform for each of the possible loan benchmarks, (i(benchmark). These are relative spreads and are most meaningfully interpreted by taking a difference between any two.BenchmarkACE transformPrime Rate17.53Libor-123.47Certificate of Deposit-133.18Cost of Funds-97.48Treasury Bills-115.91Fed Funds-158.99Bankers Acceptance-163.29 Table  SEQ Table \* ARABIC 5 Impact of Multiple Bank Lenders on Spreads This table analyzes cases in which the same borrower makes more than one loan against the same benchmark on two different dates. These are shown in the aggregate (column 3) and separated according to whether the borrower changes banks or stays with the same bank (columns 1-2). The cases are separated according to whether the spread changes or not; cases in which spread changes are then further separated between those in which (unadjusted) spreads rise and those in which they fall. Line B shows that borrowers which stay with the same bank are more than twice as likely to keep the same spread. Line C shows that, conditional on spreads changing, the likelihood of a rise or a fall is not statistically different between companies which stay with their bank and those which change banks. Borrowers withBorrowers with Total: one Banktwo or more BanksAll BorrowersA. Total cases8,1703,36811,538B. Cases in which spreads are unchanged4,3468335,179 B / A = Proportion unchanged 0.5320.2470.449Standard deviation 0.0060.0070.005t-statistic for a difference of means 15.050-27.108C. Cases in which spreads change3,8242,5356,359 C1. Spreads rise1,6721,1562,828 C2. Spreads fall2,1521,3793,531C1/C = Proportion of positive changes0.4370.4560.445Standard deviation 0.0080.0100.006t-statistic for a difference of means -0.9331.141 Table  SEQ Table \* ARABIC 6 Results of OLS Regression of Adjusted Spreads This table reports results designed to check the robustness of the ACE procedure in identifying the Prime premium. Adjusted spreads are regressed on a constant and just four variables, one of which is a Prime dummy. The R2 is 0.552.VariableCoefficientStandard Errort-statisticConstant98.1021.45967.26Risk1.7520.02377.73Log sales-25.1520.343-73.36Signing date0.8830.01656.55Prime indicator144.3780.891162.00 Table  SEQ Table \* ARABIC 7 Mean Spreads by Size and Risk of Borrower Loans are segregated into Prime-based (Panel A) and market index based (Panel B) and divided into a 4x4 matrix of risk and size ranges. Spreads are adjusted to Libor equivalents and averaged in each cell. Panel C displays the difference in means with t-values for a test of difference of means in parentheses. The difference is significant at the 99% level in all cells. Panel A: Mean Spreads (Libor-Equivalent) of Prime-Based LoansRisk: SizeR(2525J>ŽAA±A²A¹AŗAüDE+F,F~OOžOŸORSuv}~•••• • • • •••k•l••€••śöńöńöńöńöńöéöéöéöńößößößöéöéöéöŚößöÓöÓöéöéöéöĢöžŚöŚö·öéö¤˜jvĻ4 hÄEHä’U$jvĻ4 hÄCJUVmHnHu ješhÄ jbšhÄ jSšhÄ jašhÄ j£šhÄ hÄH*jhÄ0JUjhÄU hÄH*hÄ hÄCJ<*+9R_`apqrstuvwx+ , - / < }żżżżżńńńńńńńļļļļļļļļįļļļßļ „Š„Š„]„Š^„Š`„ „°„¤^„°`„haaźgżżż}ן–éL«("‰$&˜'­)O+.=1£2ŗ2m4Ż5^77*9ū;¬=/?ŽA•BėC½Dżżżżżżżżżżżżżżżżūżżżłżżżżżżżż½D E‹E]G9IK«KsMOƒPæRįTāT6VpXÄX×[õ[Ć\¼_b*bc×ekg|gdjż÷żżżõżżżżżżēē÷żõżżżõżżżõż „Š„Š„ ]„Š^„Š`„ „`„djqjl«l{n.pHpcrtrZtct†v y,yI{X{,~F~¤€ĘƒƒF…T…’†^‰ĀŠ$Œ-ŒŁŽżūżūūżūżūżūūżūżūżūūūżūżūūūūżūŁŽ’Ņ”k•‡•Å—(šDšœ(RŸ”±£§„6¦:©6Ŗ ¬.®s° ²¦³#µżūżżżģężŲżżżżżżżżżżżżżżŅ Ę* ĘĄ!B„p„^„p`„„`„ ĘĄ!$„p„^„p`„•‚•Æ•°•Ó•Ō•ą•į•ā•ä•å•]–^–ó–ō–F—G—H—J—K—Ļ˜Š˜ß˜ą˜A™B™I™P™Č™É™Ź™Ģ™Ķ™(š)š<š=š>š?šUšVšišjškšlš”š•š–š˜š™š0›1›é›ź›Zœ[œ÷óīóēóąīóīóŁóēóąīóīóĻóēóēóŹóąīóīó÷ó·«÷ó÷ó˜Œ÷óąīóīóīóŁóąjN-ģ4 hÄEHö’U$jN-ģ4 hÄCJUVmHnHujuĻ4 hÄEHä’U$juĻ4 hÄCJUVmHnHu hÄ6jhÄ0JU ješhÄ jjšhÄ jqšhÄ hÄH*hÄjhÄU7[œ\œ^œ_œ”•©Ŗ«­®ĻŠćäåęBžCžŃžŅžÓžÕžÖž$Ÿ%ŸGŸHŸOŸPŸé ź  ” ”””QÄRÄtÄuÄ|Ä}Ä÷ÉųÉŹŹ"Ź#ŹĶ$ĶYŠZŠ|Š}Š„Š…ŠT×\×MŚNŚpŚqŚxŚyŚ&Ü'ÜIÜJÜQÜRÜĮęĖęēē<ē=ēśöśöśöļśöśöēöŌČēöĆöļśöśöēöēöēöēöēöēöēöēöēöēöēöēö¾öēöēöēö¾öēöēöēöēöēöēö¾öēöē hÄ6 hÄH*j-Ö÷4 hÄEHö’U$j-Ö÷4 hÄCJUVmHnHujhÄU jjšhÄhÄ hÄH*K#µ·Įøęøs»b½Tæ,Ā2Ć#Å®ĘŪÉāĢXĪ¤ŃoÓÕÜÖ|Ų“ŲbŁqŪHŻŪŽ•ąąįä‚åĘålčżżūżżżżżżżżżżżżżżżūżżżżżżżżłż=ēDēEēéé#é$é+é,éhīwī<ļ=ļ_ļ`ļgļhļ£ļĻļ'õ(õJõKõRõSõ&'IJRSopuvwxy”¢µ¶·øÅĘ%&9:;<GüōüōüōüōüļüōüōüōüļüōüōüōüōüōüōüčüįüŚüÓŹüōü·«ōüįüōü˜Œōüj±Ļ4 hÄEHš’U$j±Ļ4 hÄCJUVmHnHuj²Ļ4 hÄEHš’U$j²Ļ4 hÄCJUVmHnHuhÄH*OJQJ jsšhÄ jhšhÄ jdšhÄ jlšhÄ hÄ6jhÄUhÄ6lč“ź ģ«ī\š+óōeōöc÷žųLś;üŃżÅ”¹%>GYƒ0 Q żżżżżżūżżżżżżżżżöšźšį׿öū Ęš„ ^„ „düž`„„`„„ ^„ düžGHMNOPpq}~‹Œ®ÆĒČüżÅ Ģ "*-*„*”*+/+ƒ+”+,&,Ŗ,Ē,N-}-Ū-õ-s..µ.Õ.X/j/Ė/é/1,1ž1°1(2:2Š2ć2^3|3ģ3444Ī4×4ł56Ś6ę67 7Ī8×8D:J:K:`:a:łõīēāõłõēõīõēõīõłõŻõÖõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŻõŃõŃõŃõŃõŹõŃõĆøĆøjhÄ5>*U hÄ5>* jjšhÄ hÄ5 hÄ5CJ hÄ6 hÄH* jsšhÄ jlšhÄhÄ jmšhÄHQ 8 ū 4ĄD°Ÿ¬»6!d"¤$i&3(!*-*­*=+®+2,Õ,‹-..ō.żżżżżżżżżżżūżżżżżżżłżżżżżżżżżō.x/÷/š071Ą1J2ó2Š3444Ķ4Ī4×4ų5ł56Ł6Ś6ę6Ķ8Ī8×8A:C:D:żżżżżżżżżżūżöūżżūżöūżöūżżżdpža:b:c:d:«:™;š;œ;¢;«;“;Ą;Ė;Ķ;Ū;Q<i<ę<ē<é<ļ<š<====,=Ž>ß>7?[?c?{?Į?Õ?÷@AÄAŹAĖAąAįAāAćAäA BkBlBmBoBpB|C}CÄCÅCĘCČCÉCÖCķCīCDDDD2DóčįÜŲÜŲÜŲÜŲÜŲÜŲÜŲŅŲÜČÜČ½ČÜŲÜŲÜŲÜŲÜŲÜŲÜČÜČ½ČܶŲÆŖŲŖŲ„ŲÆŖŲŖŲÜŲÜŲÜŲÜ hÄH* hÄH* jjšhÄ hÄ>*CJhÄ>*mHnHujhÄ>*U hÄCJhÄ hÄ>* hÄ5>*jhÄ5>*UhÄ5>*mHnHuAD:d:e:«:™;š;›;œ;;¢;µ;Ģ;ššįĶšš‹|o|| $„$If`„a$$„¤$If`„a$$„¤$If`„a$2kd$$If–lÖdøöÖ’Ö’Ö’Ö’4Ö laöŠ„°„°„¤$If]„°^„°`„$„¤š$If`„a$$„¤$If`„a$ Ģ;Ķ;Ü;Ż;Ž;ß;‘‚sss$„¤$If`„a$$„¤$If`„a$nkdN$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠß;ą;š;÷;ż;<‘‚sss$„¤$If`„a$$„¤$If`„a$nkdń$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠ<<<<%<+<‘‚sss$„¤$If`„a$$„¤$If`„a$nkd†$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠ+<,<><D<J<P<‘„uuu$„¤$If`„a$ $„$If`„a$nkd$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠP<Q<j<k<l<m<‘‚sss$„¤$If`„a$$„¤$If`„a$nkd°$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠm<n<ƒ<Š<<–<‘‚sss$„¤$If`„a$$„¤$If`„a$nkdE$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠ–<—<¬<³<¹<æ<‘‚sss$„¤$If`„a$$„¤$If`„a$nkdŚ$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠæ<Ą<Ņ<Ł<ß<å<‘‚sss$„¤$If`„a$$„¤$If`„a$nkdo$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠå<ę<ē<é< =-=Ü>Ż>‘ˆ†ylXX„Š„Š„¤$If]„Š^„Š`„ $„$If`„a$ $„$If`„a$$„p^„pa$nkd$$If–lÖ\dJ ŒĪę BBBöö Ö’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠŻ>Ž>ß>ą>?$?7?8?Ź»¬¬¬¬>nkdõ$$If–lÖ\”’l “ęp#Ų H2ŠööÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laö$„¤$If`„a$$„¤$If`„a$5kd§$$If–lÖ”’p#Ü#ööÖ’Ö’Ö’Ö’4Ö laö8?9?B?C?D?L?M?U?V?[?n?o?t?u?z?šįįššššįįšįįįį$„¤$If`„a$$„¤$If`„a$z?{?”?•?œ?¤?«?ZKK<<<$„¤$If`„a$$„¤$If`„a$¤kd½$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö«?²?¹?Ą?Į?Ö?×?šššK<š$„¤$If`„a$¤kd“$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$×?Ų?Ł?Ś?Ū?Ü?Ż?šššššK¤kd$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$Ż?ļ?ō?ū?@@ @@šįįįįįį$„¤$If`„a$$„¤$If`„a$@@,@2@7@<@A@ZK<<<<$„¤$If`„a$$„¤$If`„a$¤kd† $$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laöA@F@K@L@a@f@l@ššK<šš$„¤$If`„a$¤kdo $$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$l@r@x@~@„@…@œ@ššššK<$„¤$If`„a$¤kdX $$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$œ@”@¦@«@°@µ@ŗ@»@ššššššK¤kdA $$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$»@Ó@Ō@Ś@ß@å@ź@š@ö@ššįįįįįį$„¤$If`„a$$„¤$If`„a$ö@÷@AAAAAZK<<<<$„¤$If`„a$$„¤$If`„a$¤kd* $$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laöAAAA A%A*AššK<šš$„¤$If`„a$¤kd$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$*A/A4A9A>A?AJAššššK<$„¤$If`„a$¤kdü$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$JAOATAZA`AfAlAmAššššššK¤kdå$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$mAyA~AƒAˆAA’A—Ašįįįįįį$„¤$If`„a$$„¤$If`„a$—A˜A¢A§A¬A±A¶AZK<<<<$„¤$If`„a$$„¤$If`„a$¤kdĪ$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö¶A»AĄAĮAĀAÄAäAššKID7 $„$If`„a$dpž¤kd·$$If–l֞”’l Ć 6M“ęw#Ų Wsg0‘öÖ’’’’’’’Ö’’’’’’’Ö’’’’’’’Ö’’’’’’’4Ö laö$„¤$If`„a$äA B›CœCCÖC×Cöä±¢•†$„¤$If`„a$ $„$If`„a$$„¤$If`„a$2kd®$$If–lÖ“l!² öÖ’Ö’Ö’Ö’4Ö laö „Š„Š„$If]„Š^„Š`„ $$Ifa$×CŲCŁCīCļC÷CųCDDD1D¦—ˆyyyyyyy$„¤$If`„a$$„¤$If`„a$$„¤$If`„a$Xkdü$$If–lÖF“Œ {l!Ų īģöÖ ’’’Ö ’’’Ö ’’’Ö ’’’4Ö laö  1D2DDDKDQDWD]D€qbbbb$„¤$If`„a$$„¤$If`„a$~kd®$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö 2D]D^DE1E»EŚEŁFßFąFõFöF÷FųFłFGÆG°G±G/HHHSHYH`HhH€HˆH—HžH®H¶HĮHÉHŻHåHéHļHšHIIII I3I~LØL½LŃL MĮMŠM×MłM N\NOOOOOO1O2O3O4O5OcOBPü÷ü÷ü÷ü÷ķ÷ķāķü÷üŪÖü÷üĪüĪüĪüĪüĪüĪüĪü÷ķ÷ķāķüÄĪÄĪÄĪø«øĪÄĪø«Īü÷ķ÷ķāķü÷ühÄ56B*CJphhÄ6B*CJphhÄ>*B*phhÄB*ph hÄH* jjšhÄhÄ>*mHnHujhÄ>*U hÄ>*hÄC]D^DvD}DƒD‰DD€qbbbb$„¤$If`„a$$„¤$If`„a$~kd~$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö DDD„D«D±D·D€qbbbb$„¤$If`„a$$„¤$If`„a$~kd@$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ·DøDĻDÕDŚDßDäD€qbbbb$„¤$If`„a$$„¤$If`„a$~kd$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö äDåDżDžDE EEE€qqbbbb$„¤$If`„a$$„¤$If`„a$~kdÄ$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö EE1E2E3E4E5E6E€qqbbbb$„¤$If`„a$$„¤$If`„a$~kd†$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö 6E7E?EEEJEOETE€qbbbb$„¤$If`„a$$„¤$If`„a$~kdH$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö TEUE`EgElEqEwE€qbbbb$„¤$If`„a$$„¤$If`„a$~kd $$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö wExE„EŠEE”E™E€qbbbb$„¤$If`„a$$„¤$If`„a$~kdĢ$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ™EšE¤E„E«E°EµEŗE€qqbbbb$„¤$If`„a$$„¤$If`„a$~kdŽ$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ŗE»EŪEÜEŻEŽEßE€qbbbb$„¤$If`„a$$„¤$If`„a$~kdP$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ßEąEźEņEųEžEF“„uuuu$„¤$If`„a$$„¤$If`„a$kkd$$If–lÖ\“-'{l!yłTļöÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laö FFFFF"F(F€qbbbb$„¤$If`„a$$„¤$If`„a$~kd¾$$If–lÖr“Œ -'{l!Ų ”śTķöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö (F)F6F=FCFHFMF€qbbbb$„¤$If`„a$$„¤$If`„a$~kd€$$If–lÖr“Œ -'{l!Ų ”łTńöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö MFNF[FbFiFoFuF€qbbbb$„¤$If`„a$$„¤$If`„a$~kdB$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö uFvF|FƒF‰FŽF”F€qbbbb$„¤$If`„a$$„¤$If`„a$~kd$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ”F•F™F”FØF­F³F€qbbbb$„¤$If`„a$$„¤$If`„a$~kdĘ$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ³F“F½FÄFŹFĻFÕF€qbbbb$„¤$If`„a$$„¤$If`„a$~kdˆ $$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö ÕFÖF×FŁFłFG(H€~ylff$If $„$If`„a$dpž~kdJ!$$If–lÖr“Œ -'{l!Ų ”łTļöÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laö (H)H*H+H,H-H.HĢ½½w½½Ekdh"$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ$„¤$If`„a$2kd"$$If–lÖTböÖ’Ö’Ö’Ö’4Ö laöĄ.H/H9HGHHHSHYH¹¬ŸYJ;$„¤$If`„a$$„¤$If`„a$Ekdh#$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ $„$If`„a$ $„$If`„a$EkdĢ"$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄYHZH`HhHiH€HˆH¹Ŗ›UŖ›Ekd„$$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ$„¤$If`„a$$„¤$If`„a$Ekdö#$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄˆH‰H—HžHŸH®H¶H¹Ŗ›UŖ›Ekd %$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ$„¤$If`„a$$„¤$If`„a$Ekd%$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ¶H·HĮHÉHŹHŻHåH¹Ŗ›UŖ›Ekd¼&$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄ$„¤$If`„a$$„¤$If`„a$Ekd.&$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄåHęHēHéH I4I5IQLRL¹““„––‚–„Š„Š„¤$If]„Š^„Š`„$„¤$If`„a$$$„$If`„a$dpžEkdJ'$$If–lÖ0T*bÖ 8öÖ’’Ö’’Ö’’Ö’’4Ö laöĄRLSLTLcLsL|L}L~LŹ»¬¬¬>»mkd<($$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö$„¤$If`„a$$„¤$If`„a$4kdę'$$If–”śÖā’V$s$öÖ’Ö’Ö’Ö’4Ö aö~L‡L™L§LØL©LŖL«L¬Lššš‚ssss$„¤$If`„a$mkdž($$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö$„¤$If`„a$¬L­L½LĆLÉLŠL‘…vvv$„¤$If`„a$ „¤$If`„mkd²)$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöŠLŃLśLMM M‘…vvv$„¤$If`„a$ „¤$If`„mkdf*$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö M MCMIMOMUM‘‚‚‚‚$„¤$If`„a$mkd+$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöUMVM†MŒM’M˜M‘‚‚‚‚$„¤$If`„a$mkdĪ+$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö˜M™MĮMČMŠMŃM‘‚‚‚‚$„¤$If`„a$mkd‚,$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöŃMŅMÓMŌMÕMÖM‘‚‚‚‚$„¤$If`„a$mkd6-$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöÖM×MłM’MN N‘…vvv$„¤$If`„a$ „¤$If`„mkdź-$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö N N!N'N-N3N‘…vvv$„¤$If`„a$ „¤$If`„mkdž.$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö3N4NINONUN[N‘…vvv$„¤$If`„a$ „¤$If`„mkdR/$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö[N\N„NŠNN–N‘‚‚‚‚$„¤$If`„a$mkd0$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aö–N—NĒNĶNÓNŁN‘‚‚‚‚$„¤$If`„a$mkdŗ0$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöŁNŚNO OOO‘‚‚‚‚$„¤$If`„a$mkdn1$$If–”ėÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöOOOOO5OcONP‘‚uc„Š„Š„$If]„Š^„Š`„ $„$If`„a$ $„$If`„a$mkd"2$$If–”śÖ\ā’Ņ©V$š×Ö×öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö aöBPCPOP€PQ$Q%Q:Q;Q*hÄ5>*mHnHujhÄ5>*U hÄ5>* hÄ>*hÄ hÄH*JNPOPXPdPsPP€PĢæ²²²Fkkd23$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠ $„$If`„a$ $„$If`„a$2kdä2$$If–lÖd„ öÖ’Ö’Ö’Ö’4Ö laöŠ€P‰PP–PœPP¢PØP®P“PšįįįušįįįkkdŠ3$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠ$„¤$If`„a$$„¤$If`„a$ “PµPæPĒPĶPŌP“„uuu$„¤$If`„a$$„¤$If`„a$kkd`4$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠŌPÕPāPčPīPōP“„uuu$„¤$If`„a$$„¤$If`„a$kkdš4$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠōPõPQ QQQ“„uuu$„¤$If`„a$$„¤$If`„a$kkd€5$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠQQQQ>QhQŽRßR“ŽŒp\„°„°„¤$If]„°^„°`„$„¤š$If`„a$ $„$If`„a$dpžkkd6$$If–lÖ\dō $T„ 000öÖ’’’’Ö’’’’Ö’’’’Ö’’’’4Ö laöŠßRąRS S&S+S0S8S@SESĢæŒ}naaaa $„$If`„a$$„¤$If`„a$$„¤$If`„a$2kdü6$$If–lÖv ŖöÖ’Ö’Ö’Ö’4Ö laöā $„$If`„a$2kd®6$$If–lÖv ŖöÖ’Ö’Ö’Ö’4Ö laöā ESFSMSSSYS_SeS€qbbbb$„¤$If`„a$$„¤$If`„a$~kdX7$$If–lÖrv˜ŗ Üž """""öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāeSfSoSuS{SS‡S€qbbbb$„¤$If`„a$$„¤$If`„a$~kdā7$$If–lÖrv˜ŗ Üž """""öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā‡SˆS‘S—SS£S©S€qbbbb$„¤$If`„a$$„¤$If`„a$~kdl8$$If–lÖrv˜ŗ Üž """""öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā©SŖSµS»SĮSĒSĶS€qbbbb$„¤$If`„a$$„¤$If`„a$~kdö8$$If–lÖrv˜ŗ Üž """""öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāĶSĪSĻST€~q $„$If`„a$~kd€9$$If–lÖrv˜ŗ Üž """""öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāTTTT T(T0T5T„–‡zzzz $„$If`„a$$„¤$If`„a$$„¤$If`„a$Ykd:$$If–lÖÖv ŖÖ0’’’’’’öÖ’Ö’Ö’Ö’4Ö laöā5T6T=TBTHTNTTTYJ;;;;$„¤$If`„a$$„¤$If`„a$„kd™:$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāTTUT^TdTjTpTvTYJ;;;;$„¤$If`„a$$„¤$If`„a$„kd:;$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāvTwT€T†TŒT’T˜TYJ;;;;$„¤$If`„a$$„¤$If`„a$„kdŪ;$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā˜T™T¤TŖT°T¶T¼TYJ;;;;$„¤$If`„a$$„¤$If`„a$„kd|<$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā¼T½T¾TųTYWJ $„$If`„a$„kd=$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāųTłT’TU UUUU„–‡zzzz $„$If`„a$$„¤$If`„a$$„¤$If`„a$Ykd¾=$$If–lÖÖv ŖÖ0’’’’’’öÖ’Ö’Ö’Ö’4Ö laöāUU&U,U3UYJ;,$„¤$If`„a$$„¤$If`„a$$„¤$If`„a$„kd?>$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā3U9U@UFUMUSUZUšįšįšį$„¤$If`„a$$„¤$If`„a$ZU[UdUiUpUYJ;,$„¤$If`„a$$„¤$If`„a$$„¤$If`„a$„kdą>$$If–lÖÖrv˜Ø Üž "4""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāpUvU}UƒUŠUU—Ušįšįšį$„¤$If`„a$$„¤$If`„a$—U˜U”U§U®UYJ;,$„¤$If`„a$$„¤$If`„a$$„¤$If`„a$„kd«?$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöā®U“U»UĮUČUĪUÕUšįšįšį$„¤$If`„a$$„¤$If`„a$ĪUŌUŽUßUēUķUōUśUVVVVVV V5V6V7V8VhVēVéVXDXEXLXMXTXUXdXX€XŒXX‘X’XšX›X³X“XĄXĮXÅXĘXĻXŠXčXéXõXöXūXüXYYYY*Y+Y0Y1Y:Y;YHYIYNYOYPYQY^Y_YdYeYfYgYtYuYųōķōųōųōųōųōčŽčŽÓŽčōĪōčĘčĘčĘčō¾ō¾ō¾ōķō¾ō¾ō¾ōķō¾ō¾ō¾ōķō¾ō¾ō¾ō¾ō¾ō¾ō¾ō¾ō¾ō¾ō¾jhÄU j£šhÄ>* hÄH*hÄ>*mHnHujhÄ>*U hÄ>* j£šhÄhÄ hÄ6CJKÕUÖUįUēUīUYJ;,$„¤$If`„a$$„¤$If`„a$$„¤$If`„a$„kdL@$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāīUōUūUVVVVšįšįšį$„¤$If`„a$$„¤$If`„a$VVVV9VhVYWWJ;$„¤š$If`„a$ $„$If`„a$„kdķ@$$If–lÖÖrv˜ŗ Üž """""Ö0’’’’’’öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöāhVXX X@XAXBXCXHXPXXXķą­žąką^^^ $„$If`„a$2kdÜA$$If–lÖh !¤öÖ’Ö’Ö’Ö’4Ö laöŌ$„¤$If`„a$2kdŽA$$If–lÖh !¤öÖ’Ö’Ö’Ö’4Ö laöŌ $„$If`„a$„Š„Š„$If]„Š^„Š`„ XX]XcXdXkXpXuXzXņņ`QBBB$„¤$If`„a$$„¤$If`„a$‘kd8B$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ $„$If`„a$zXX“X”XX¢XØX®X³XČXšš^Oššššš$„¤$If`„a$‘kdęB$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ$„¤$If`„a$ ČXÉXŅX'$„¤$If`„a$Ųkd”C$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽ Ö Ö<’’’’’’’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌpÖ<’’’’’’’’’’’’ŅX×XŻXćXčXżXššššš$„¤$If`„a$żXžX Y'$„¤$If`„a$ŲkdŻD$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽ Ö Ö<’’’’’’’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌpÖ<’’’’’’’’’’’’ YYYYY2Y3Y:Yššššš^M$„¤x¤$If`„a$‘kd&F$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ$„¤$If`„a$:YPYfY|Y’Y©YŖY«YćYīīīīī\ZK$„¤$If`„a$‘kdŌF$$If–lֈhĖ¤ }V/ !cŁŁŁŁŽöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ$„¤x¤$If`„a$uYzY{Y|Y}YŠY‹YY‘Y’Y“Y Y”Y¤Y„Y«YēYčYļYšY÷YųYZ$Z%Z1Z2Z7Z8Z@ZAZXZYZeZfZkZlZtZuZ‹ZŒZ˜Z™ZZžZØZ©ZæZĄZĢZĶZŃZŅZŪZÜZéZźZļZšZńZņZ’Z[[[[[[[[[[[+[,[0[1[=[C[D[Y[Z[üōüōüōüōüōüōüōüļēļēļēļüōüōüōüąüōüōüōüąüōüōüōüąüōüōüōüōüōüōüōüōüōüōüōüōüōüōüōüŁĪŁĪjhÄ5>*U hÄ5>* j£šhÄ j£šhÄ>* hÄ>*jhÄUhÄQćYäYåYęYėYóYūYZZņ˜ņ‹‹‹‹‹ $„$If`„a$YkdG$$If–lÖÖh !¤Ö0’’’’’’öÖ’Ö’Ö’Ö’4Ö laöŌ $„$If`„a$ZZZZZF7(($„¤$If`„a$$„¤$If`„a$økdH$$If–lÖֈhĖ¤ }V/ !cŁŁŁŁŽÖ0’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌZZ$Z9Z:Zššš7økdÖH$$If–lÖֈhĖ¤ }V/ !c’’’’Ł’’’’Ł’’’’Ł’’’’Ł’’’’Ž’’’’Ö0’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’’’’’’’’’’’’’’’’’’’Ö’’’’’’4Ö laöŌ$„¤$If`„a$:ZCZIZNZSZXZmZnZwZ|ZZ†Z‹ZŸZ Z«Z°ZµZŗZæZÓZšįįįįįÜšįįįįįךįįįįįFf‚NFfK$„¤$If`„a$$„¤$If`„a$ÓZŌZŪZńZ[F5$$$„¤x¤$If`„a$$„¤x¤$If`„a$økdP$$If–lÖֈhĖ¤ }V/ !c’’’’Ł’’’’Ł’’’’Ł’’’’Ł’’’’Ž’’’’Ö0’’’’’’öÖ’’’’’’’’’’’’’’’’’’’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ[[2[9[:[<[īīī53økdRQ$$If–lÖֈhĖ¤ }V/ !cŁŁŁŁŽÖ0’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ$„¤x¤$If`„a$<[=[][œ[‹\Œ\\–\—\Ÿ\ \Ø\©\®\Å\żššŽ«œš€€€€ $„$If`„a$$„¤$If`„a$$„¤$If`„a$2kdR$$If–lÖh !¤öÖ’Ö’Ö’Ö’4Ö laöŌ„Š„Š„$If]„Š^„Š`„ $„$If`„a$Z[[[\[][œ[Œ\Ę\]]m]s]¢]Ø]©]¾]æ]Į]Ā]ē]£_`5`C`F`M`y`‡`Š`‘`»`É`Ģ`Ó`ž` aaa4a>a@aUadahaiaoapaqasataza{a|a~aa…a†aˆa‰aŠaaŽa™bšbóčäßäßäßäßäŲčŲčóčßäßäŃäŃäŃäŃäŃäŃäŃäŃäĢäŃĆä¹³¹³ä¹³¹³ä¹³¹Ø¹³äžäžjhÄ0JUhŲ5…0JmHnHu hÄ0JjhÄ0JUhÄ56CJ hÄ5 hÄ6CJ hÄ5>* hÄ>*hÄjhÄ5>*UhÄ5>*mHnHu>Å\Ę\×\Ž\ä\é\ī\€qbbbb$„¤$If`„a$$„¤$If`„a$~kdeR$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌī\ļ\] ]]]]€qbbbb$„¤$If`„a$$„¤$If`„a$~kd5S$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌ]]4]9]?]E]J]€qbbbb$„¤$If`„a$$„¤$If`„a$~kd÷S$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌJ]K]a]g]m]s]x]€qbbbb$„¤$If`„a$$„¤$If`„a$~kd¹T$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌx]y]‡]Ž]”]š]Ÿ]€qbbbb$„¤$If`„a$$„¤$If`„a$~kd{U$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌŸ] ]”]¢]Ć]ē]¤_€~~qbN„Š„Š„¤š$If]„Š^„Š`„$„¤š$If`„a$ $„$If`„a$~kd=V$$If–lÖrh¬ ©Kž !D ż¢³öÖ’’’’’Ö’’’’’Ö’’’’’Ö’’’’’4Ö laöŌ¤_„_¦_§_Æ_µ_Ą_Ę_Š_Ū_ė_õ_``Ģ½½½®”®®®®®®® $„$If`„a$$„¤$If`„a$$„¤$If`„a$2kd W$$If–lÖh !¤öÖ’Ö’Ö’Ö’4Ö laöŌ ````#`)`.`3`m^OOOOO$„¤$If`„a$$„¤$If`„a$‘kd[W$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ3`4`5`<`C`F`M`P`m^^^^^^$„¤$If`„a$‘kdX$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌP`Q`Y`_`e`k`p`w`m^OOOOO$„¤$If`„a$$„¤$If`„a$‘kdÅX$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌw`x`y`€`‡`Š`‘`”`m^^^^^^$„¤$If`„a$‘kdsY$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ”`•``£`©`Æ`“`¹`m^OOOOO$„¤$If`„a$$„¤$If`„a$‘kd!Z$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ¹`ŗ`»`Ā`É`Ģ`Ó`Ö`m^^^^^^$„¤$If`„a$‘kdĻZ$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌÖ`×`ß`å`ė`ń`ö`ü`m^OOOOO$„¤$If`„a$$„¤$If`„a$‘kd}[$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌü`ż`ž`a aaaam^^^^^^$„¤$If`„a$‘kd+\$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌaa"a(a.a4a9a>am\KKKKK$„¤x¤$If`„a$$„¤x¤$If`„a$‘kdŁ\$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ>a?a@aGaNaUa\adam^^^^^^$„¤$If`„a$‘kd‡]$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌdaeafahaqarasa|a}a~amkkb\kSQk„ü’„&`#$„h]„h„ų’„&`#$‘kd5^$$If–lֈh+ ;K[ !³ÆöÖ’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌ ~aŠa‹aŒaa™bšc_eŹeégźgėgöōņņšššššņņ„ü’„&`#$ šbšc›c_e`eŹeĖeėgüņüņüņüjhÄ0JUhÄ) 000 BP°Š/ °ą=!°"°# $ %°L$$If–Š!vh5Öø#vø:V –löö5Öø4ÖaöŠ”$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö  ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ“$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö ’4ÖaöŠ”$$If–Š!vh5Öę 5ÖB5ÖB5ÖB#vę #vB:V –lööö 5Öę 5ÖB/Ö ’/Ö  ’/Ö ’4ÖaöŠL$$If–!vh5ÖÜ##vÜ#:V –lööö5ÖÜ#4ÖĘ$$If–!vh5ÖŲ 5ÖH5Ö25ÖŠ#vŲ #vH#v2#vŠ:V –lööö5ÖŲ 5ÖH5Ö25ÖŠ/Ö ’/Ö  ’/Ö ’/Ö ’4Öõ$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö  ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5Ö  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~€‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸ ”¢£¤„¦§Ø©Ŗ«¬­®Æ°±²³“µ¶·ø¹ŗ»¼½¾æĄĮĀĆÄÅĘĒČÉŹĖĢĶĪĻŠŃŅÓŌÕÖ×ŲŁŚŪÜŻŽßąįāćäåęēčéźėģķīļšńņóōõö÷ųłśūüżž’      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNž’’’PQRSTUVxż’’’ż’’’ż’’’[` ^_baecdfghijklmnpoqrsvtu”¢ż’’’yz{|}~€Root EntryO<@ł <@ō é <@ō ’’’’’’’’/ ĄF@FcžįXĀ]@#Data… <@ō Key1exž_š-5.70ü’ ’’’’’’’’’’’’I <@ō Oń^WordDocumentå <@ō .’’’’’’’’š3.21ü’¹ <@ō © +œObjectPool1’’’’ ė0žįXĀ@FcžįXĀ-1.13ü’_885985398 ĄF ė0žįXĀĄŒ8žįXĀOle ’’’’’’’’’’’’PIC ’’’’LMETA ’’’’’’’’’’’’Øž’’’ž’’’ ž’’’ž’’’ž’’’ž’’’ž’’’ž’’’ !"#$%&'()*ž’’’,ž’’’ž’’’/0ž’’’ž’’’3ž’’’56789ž’’’;ž’’’ž’’’ž’’’ž’’’@ž’’’BCDEFž’’’Hž’’’ž’’’ž’’’ž’’’Mž’’’OPQRSTUVWXYZ[\]ž’’’_ž’’’ž’’’bcž’’’ž’’’fž’’’hijklmnopqrstuvwž’’’yž’’’ž’’’|}ž’’’€LµÆčØččµÆ5 ½ ’’’.1  @@&’’’’Ą’ŗ’ś & MathTypeąū€žSymbol- 2 €0qĒ 2 €‹ję 2 €W e„ū€žTimes New Roman-š 2 €ü(~ 2 €X)~ 2 €ė(~ 2 €“ )~ 2 €’y“ 2 € xĆū ’Times New Roman-š 2 ąƒi> 2 ąL i> 2 õāi> 2 īFnpū€žSymbol-š 2 €Ś=Ó 2 €Ż +Óū ’Symbol-š 2 õ6={ūĄżSymbol-š 2 ×®å˜ū ’Times New Roman-š 2 õ·1p & ’’’’ū¼"System-šŒŠČŽŲĒ}6hÖ hĪ hŽ hB6jž’ ’’’’ĄFMicrosoft Equation 2.0 DS Equation Equation.2Ņ€G?ŌO>d O> „q‚(ƒy‚)˜˜˜†=˜˜˜„j ƒi ‚(ƒx ƒi ‚)˜˜†+˜˜„e ƒi†=ˆ1ƒn †åolCompObj ’’’’ZObjInfo’’’’ ’’’’Equation Native ’’’’’’’’’’’’œ_885985397’’’’’’’’ ĄFĄŒ8žįXĀĄŒ8žįXĀOle ’’’’’’’’’’’’PIC  ’’’’LMETA ’’’’’’’’’’’’čCompObj’’’’+ZL‰Æ` Øčč‰Æ5 ē ’’’.1  @&’’’’Ą’ŗ’Ąś & MathTypeąś-–’–Fū€žTimes New Roman2~- 2 €@y“ 2 €’y“ 2 €4 xĆū ’Times New Roman-š 2 ą i> 2 ą i> 2 õyi> 2 īŁnpū€žSymbol-š 2 €Ś=Ó 2 €ß+Ó 2 €¤ +Óū ’Symbol-š 2 õÉ={ūĄżSymbol-š 2 ×Aå˜ū€žSymbol-š 2 €ję 2 €e„ū€žTimes New Roman2~-š 2 €ž (~ 2 €“ )~ū ’Times New Roman-š 2 õF1p & ’’’’ū¼"System-š7€/ž’ ’’’’ĄFMicrosoft Equation 2.0 DS Equation Equation.21Ķ o2ä1t1 ƒy˜˜˜†=˜˜˜2ƒy˜˜†+˜˜„j ƒi ‚(ƒx ƒi ‚) ƒObjInfo’’’’’’’’-Equation Native ’’’’’’’’’’’’.¼_887893326 ĄFĄŒ8žįXĀĄŒ8žįXĀOle ’’’’’’’’’’’’1i†=ˆ1ƒn †å ˜˜†+˜˜„eDĪ$s4/L`Č,čč`<   ’’’.1  ą@&’’’’Ą’ą’Ą & MathTypePś-V@Vōū€žTimes New Roman- 2 @@y“ PIC ’’’’2LMETA ’’’’’’’’’’’’4hCompObj’’’’:ZObjInfo’’’’’’’’<& ’’’’ū¼"System-šŻŻŻĄŻŻŻŻŻŻŻŻŻŻŻĄ»»»»»»»»»»» »»»»ž’ ’’’’ĄFMicrosoft Equation 2.0 DS Equation Equation.21Ķ o2l1Ø1 2ƒy L`ķČččEquation Native ’’’’’’’’’’’’=<_888657453’’’’’’’’ĄFĄŒ8žįXĀĄŒ8žįXĀOle ’’’’’’’’’’’’>PIC ’’’’?L`ķv2   ’’’.1  Ą@&’’’’Ą’č’Ø & MathTypePś-J@Jåū žTimes New Romang8- 2 @y„ & ’’’’ū¼"System-š÷W{’÷’ļu’÷’×>{w÷÷ouw÷÷Wt{’÷’ž’ ’’’’ĄFMicrosoft Equation 2.0 DS EqMETA ’’’’’’’’’’’’AhCompObj’’’’GZObjInfo’’’’’’’’IEquation Native ’’’’’’’’’’’’J<uation Equation.21Ķ o2l1Ø1 2ƒy L9Ä øčč95 å ’’’.1  Ą &’’’’Ą’³’`s &_885986994'’’’’"ĄFĄŒ8žįXĀĄŒ8žįXĀOle ’’’’’’’’’’’’KPIC !$’’’’LLMETA ’’’’’’’’’’’’Nč MathType€ūkżćSymbol„-2 ![ūkżćSymbol„-š2 M]ūkżćSymbol„-š2 x [ūkżćSymbol„-š2 č]ū€žTimes New Roman-š 2 Ą@Eź 2 Ą„eØ 2 ĄĮeØ 2 Ą— Eź 2 Ąü eØū ’Times New Roman-š 2 €WŌ 2 × WŌū ’Symbol-š 2 s-{ 2 -{ 2 Ź -{ū€žSymbol-š 2 Ą‘=Óū ’Symbol-š 2 l} 2 ld}k 2 Z l} & ’’’’ū¼"System-š7€/ž’ ’’’’ĄFMicrosoft Equation 2.0 DS Equation Equation.2CompObj#%’’’’^ZObjInfo’’’’&’’’’`Equation Native ’’’’’’’’’’’’a¼_885986993’’’’’’’’)ĄFĄŒ8žįXĀĄŒ8žįXĀ1Ķ o2,1¼1 ƒEƒe †-„lƒW ˜ƒe †-„l„d –[–]˜˜˜†=˜˜˜˜ƒEƒe †-„lƒW –[–]„eDĪ$s4/Lš+Œ Ģččš+5  ’’’.Ole ’’’’’’’’’’’’dPIC (+’’’’eLMETA ’’’’’’’’’’’’g(CompObj*,’’’’xZ1  ąą&’’’’Ą’Ø’ ˆ & MathType€ūkżćSymbol„-2 )o[ūkżćSymbol„-š2 )]ū€žTimes New Roman-š 2 ąŽEź 2 ąóeØ 2 ąy eØū ’Symbol-š 2 4Į-{ 2 4G -{ 2 4\ -{ū€žSymbol-š 2 ąĆ=Ó 2 ąĶ=Óū ’Symbol-š 2 4Qld}k 2 4× l} 2 4Å m} 2 4č ls}†ū€žTimes New Roman-š 2 ąW1Ąū`’Times New Roman-š 2 Šż2Pū ’Times New Roman-š 2 4Ī2p 2 4h (I 2 4|/> 2 4N)I & ’’’’ū¼"System-šTimes New Romanž’ ’’’’ĄFMicrosoft Equation 2.0 DS Equation Equation.21Ķ o2`1$1 ˜ƒEƒe †-„l„d –[–]˜˜˜†=˜˜˜ˆ1˜˜˜†=˜˜˜ƒe †-„l‚(„m†-„l„s ˆ2 ‚/ˆ2‚)4/ž’ ą…ŸņłOh«‘+'³Ł0œ˜ČŌšü $0ObjInfo’’’’-’’’’zEquation Native ’’’’’’’’’’’’{¼1Table’’’’¤”ŽSummaryInformation(0’’’’~Ģ‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸž’’’ž’’’£ž’’’ž’’’„¦§Ø©Ŗ«¬­®Æ°±²³“µ¶·ø¹ŗ»¼½¾æĄĮĀĆÄÅĘĒČÉŹĖĢĶĪĻŠŃŅÓŌÕÖ×ŲŁŚŪÜŻŽßąįāćäåęēčéźėģķīļšńņóōõö÷ųłśūüžż’’’’Ų 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öē$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö ’4Öõ$$If–!vh5ÖŲ 5ÖW5Ös5Ö5Ög5Ö25Ö‘#vŲ #vW#vs#v#vg#v2#v‘:V –löö5ÖŲ 5ÖW5Ös5Ö5Ög5Ö05Ö‘/Ö ’/Ö  ’/Ö ’4ÖL$$If– !vh5Öø #vø :V –löö5Ö² 4Öaö °$$If– !vh5ÖŲ 5Öļ5Öń#vŲ #vļ#vń:V –löö5ÖŲ 5Öī5Öģ/Ö ’/Ö  ’/Ö ’/Ö ’4Öaö Ī$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö  ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ŗ$$If– !vh5Öy5Öś5ÖT5Öń#vy#vś#vT#vń:V –löö5Öy5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öś5ÖT5Öķ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öń/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ą$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö ’4Öaö Ī$$If– !vh5ÖŲ 5Ö”5Öś5ÖT5Öń#vŲ #v”#vś#vT#vń:V –löö5ÖŲ 5Ö”5Öł5ÖT5Öļ/Ö ’/Ö  ’/Ö ’4Öaö L$$If–Ą!vh5Ö#v:V –löö5Ö4ÖaöĄb$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö84ÖaöĄš$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’/Ö  ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄŒ$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö  ’/Ö ’/Ö ’4ÖaöĄš$$If–Ą!vh5ÖÖ 5Ö8#vÖ #v8:V –löö5ÖÖ 5Ö8/Ö ’/Ö ’/Ö ’/Ö  ’4ÖaöĄT$$If–!vh5Öt$#vt$:V –”śöö5Ös$4Ö4Ö Ą$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö  ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö ²$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”ėöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö ’4Ö4Ö Ą$$If–!vh5Öš5Ö×5ÖÖ5Ö×#vš#v×#vÖ#v×:V –”śöö5Öš5Ö×5ÖÖ5Ö×/Ö ’/Ö  ’/Ö ’4Ö4Ö L$$If–Š!vh5Ö #v :V –löö5Ö 4ÖaöŠœ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö  ’/Ö ’4ÖaöŠŽ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö ’4ÖaöŠŽ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö ’4ÖaöŠŽ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö ’4ÖaöŠŽ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö ’4ÖaöŠœ$$If–Š!vh5֐ 5Ö05Ö05Ö0#v #v0:V –löö5֐ 5Ö0/Ö ’/Ö  ’/Ö ’4ÖaöŠL$$If–ā!vh5ÖŖ#vŖ:V –löö5ÖŖ4ÖaöāZ$$If–ā!vh5ÖŖ#vŖ:V –löö5ÖŖ/Ö  ’4Öaöāˆ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –löö5Ö"/Ö ’/Ö ’4Öaöāˆ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –löö5Ö"/Ö ’/Ö ’4Öaöāˆ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –löö5Ö"/Ö ’/Ö ’4Öaöāˆ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –löö5Ö"/Ö ’/Ö ’4Öaöā–$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –löö5Ö"/Ö ’/Ö  ’/Ö ’4Öaöā$$If–ā!vh5ÖŖ#vŖ:V –lÖ0’’’’’’öö5ÖŖ4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4Öaöā$$If–ā!vh5ÖŖ#vŖ:V –lÖ0’’’’’’öö5ÖŖ4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāÉ$$If–ā!vh5Ö"5Ö5Ö45Ö"5Ö"#v"#v#v4#v":V –lÖ0’’’’’’öö5Ö"5Ö5Ö45Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāŸ$$If–ā!vh5Ö"5Ö"5Ö"5Ö"5Ö"#v":V –lÖ0’’’’’’öö5Ö"4ÖaöāL$$If–Ō!vh5Ö„#v„:V –löö5Ö¤4ÖaöŌZ$$If–Ō!vh5Ö„#v„:V –löö5Ö¤/Ö  ’4ÖaöŌ¬$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –löö5Öc5ÖŁ5ÖŽ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –löö5Öc5ÖŁ5ÖŽ/Ö ’/Ö ’4ÖaöŌG$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –l Ö Ö<’’’’’’’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’/Ö  ’/Ö ’4ÖaöŌpÖ<’’’’’’’’’’’’G$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –l Ö Ö<’’’’’’’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’/Ö  ’/Ö ’4ÖaöŌpÖ<’’’’’’’’’’’’¬$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –löö5Öc5ÖŁ5ÖŽ/Ö ’/Ö ’4ÖaöŌŗ$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –löö5Öc5ÖŁ5ÖŽ/Ö ’/Ö  ’/Ö ’4ÖaöŌ$$If–Ō!vh5Ö„#v„:V –lÖ0’’’’’’öö5Ö¤4ÖaöŌĆ$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –lÖ0’’’’’’öö5Öc5ÖŁ5ÖŽ4ÖaöŌŃ$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –lÖ0’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’’’’’’’’4ÖaöŌl$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –l Ö Ö<’’’’’’’’’’’’Ö0’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’/Ö ’’’’’’’’4ÖaöŌpÖ<’’’’’’’’’’’’ūkd©I$$If–lÖֈhĖ¤ }V/ !c’’’’Ł’’’’Ł’’’’Ł’’’’Ł’’’’Ž’’’’ Ö Ö<’’’’’’’’’’’’Ö0’’’’’’öÖ’’’’’’Ö’’’’’’Ö’’’’’’’’’’’’’’’’’’’’’’’’Ö’’’’’’4Ö laöŌpÖ<’’’’’’’’’’’’l$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –l Ö Ö<’’’’’’’’’’’’Ö0’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’’’’’’’’/Ö ’4ÖaöŌpÖ<’’’’’’’’’’’’ūkdM$$If–lÖֈhĖ¤ }V/ !c’’’’Ł’’’’Ł’’’’Ł’’’’Ł’’’’Ž’’’’ Ö Ö<’’’’’’’’’’’’Ö0’’’’’’öÖ’’’’’’’’’’’’’’’’’’’’’’’’Ö’’’’’’Ö’’’’’’Ö’’’’’’4Ö laöŌpÖ<’’’’’’’’’’’’Ń$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –lÖ0’’’’’’öö5Öc5ÖŁ5ÖŽ/Ö ’’’’’’’’4ÖaöŌĆ$$If–Ō!vh5Öc5ÖŁ5ÖŁ5ÖŁ5ÖŁ5ÖŽ#vc#vŁ#vŽ:V –lÖ0’’’’’’öö5Öc5ÖŁ5ÖŽ4ÖaöŌL$$If–Ō!vh5Ö¤#v¤:V –löö5Ö¤4ÖaöŌĪ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö  ’/Ö ’4ÖaöŌĄ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö ’4ÖaöŌĄ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö ’4ÖaöŌĄ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö ’4ÖaöŌĄ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö ’4ÖaöŌĪ$$If–Ō!vh5ÖD 5Öż5Ö¢5Ö³5Ö#vD #vż#v¢#v³#v:V –löö5ÖD 5Öż5Ö¢5Ö³5Ö/Ö ’/Ö  ’/Ö ’4ÖaöŌL$$If–Ō!vh5Ö¤#v¤:V –löö5Ö¤4ÖaöŌŗ$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö  ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌ¬$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö ’4ÖaöŌŗ$$If–Ō!vh5Ö³5Ö5Ö5Ö5Ö5Ö±#v³#v#v±:V –löö5Ö³5Ö5ÖÆ/Ö ’/Ö  ’/Ö ’4ÖaöŌ‚ƒ„…ž’’’‡ˆ‰Šž’’’Œž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ L X d p|„Œ”ä&The Determinants of Bank Loan Pricingofhe Network Supervisor etw Normal.dote David Beime4viMicrosoft Word 10.0@Ŗær@¤łgņĄ@ŅÆüņĄ@ŠęŪįXĀ3Ū%ž’ ÕĶ՜.“—+,ł®0  hp|„Œ” œ¤¬DocumentSummaryInformation8’’’’’’’’’’’’†<CompObj’’’’’’’’’’’’‹j’’’’’’’’’’’’’’’’’’’’’’’’“ ¼ īäer°øX­ &The Determinants of Bank Loan Pricing Titlež’ ’’’’ ĄFMicrosoft Word Document MSWordDocWord.Document.8ō9²qœ^@ń’^ Normal$$„Š¤x5$7$8$9DH$`„Ša$CJ_HmH sH tH N@N Heading 1$$„¤h@&`„a$>*KHN@N Heading 2$$„¤h@&`„a$5CJJ@J Heading 3$$„¤š@&`„a$6DA@ņ’”D Default Paragraph FontVi@ó’³V  Table Normal :V ö4Ö4Ö laö (k@ō’Į(No List FžOņF Author$„P„¤^„P`„a$CJ>*@¢> Endnote ReferenceH*8+@8  Endnote TextCJ4@"4 Header  ĘąĄ!.)@¢1. Page Number2"@2 Caption¤x54 @R4 Footer  ĘąĄ!¹/Į2I6+>Ļė_  Ņ=\_ė_œ’’’’*+9R_`apqrstuvwx+,-/<} × ź –éL«(‰˜­!O#&=)£*ŗ*m,Ż-^//*1ū3¬5/7Ž9•:ė;½< =‹=]?9AC«CsEGƒHæJįLāL6NpPÄP×SõSĆT¼WZ*Z[×]k_|_dbqbd«d{f.hHhcjtjZlcl†n q,qIsXs,vFv¤xĘy{{F}T}’~^Ā‚$„-„Ł†’‡ˆŅˆŒk‡Å(’D’”(•R—™±›§6ž:”6¢ ¤.¦sØ Ŗ¦«#­ÆĮ°ę°s³bµT·,ŗ2»#½®¾ŪĮāÄXʤÉoĖĶÜĪ|Š“ŠbŃqÓHÕŪ֕ŲąŁ܂ŻĘŻlą“ā ä«ę\č+ėģeģīcļžšLņ;ōŃõųśÅū”ü¹ü%ż>żGżYżƒ’0Q8ū4 Ą D °Ÿ¬»6d¤i3 !"-"­"=#®#2$Õ$‹%&&ō&x'÷'š(7)Ą)J*ó*Š+,,,Ķ,Ī,×,ų-ł-.Ł.Ś.ę.Ķ0Ī0×0A2C2D2d2e2«2™3š3›3œ33¢3µ3Ģ3Ķ3Ü3Ż3Ž3ß3ą3š3÷3ż34444%4+4,4>4D4J4P4Q4j4k4l4m4n4ƒ4Š44–4—4¬4³4¹4æ4Ą4Ņ4Ł4ß4å4ę4ē4é4 5-5Ü6Ż6Ž6ß6ą67$7778797B7C7D7L7M7U7V7[7n7o7t7u7z7{7”7•7œ7¤7«7²7¹7Ą7Į7Ö7×7Ų7Ł7Ś7Ū7Ü7Ż7ļ7ō7ū788 888,82878<8A8F8K8L8a8f8l8r8x8~8„8…8œ8”8¦8«8°8µ8ŗ8»8Ó8Ō8Ś8ß8å8ź8š8ö8÷899999999 9%9*9/94999>9?9J9O9T9Z9`9f9l9m9y9~9ƒ9ˆ99’9—9˜9¢9§9¬9±9¶9»9Ą9Į9Ā9Ä9ä9 :›;œ;;Ö;×;Ų;Ł;ī;ļ;÷;ų;<<<1<2<D<K<Q<W<]<^<v<}<ƒ<‰<<<<„<«<±<·<ø<Ļ<Õ<Ś<ß<ä<å<ż<ž<= ====1=2=3=4=5=6=7=?=E=J=O=T=U=`=g=l=q=w=x=„=Š==”=™=š=¤=„=«=°=µ=ŗ=»=Ū=Ü=Ż=Ž=ß=ą=ź=ņ=ų=ž=>>>>>">(>)>6>=>C>H>M>N>[>b>i>o>u>v>|>ƒ>‰>Ž>”>•>™>”>Ø>­>³>“>½>Ä>Ź>Ļ>Õ>Ö>×>Ł>ł>?(@)@*@+@,@-@.@/@9@G@H@S@Y@Z@`@h@i@€@ˆ@‰@—@ž@Ÿ@®@¶@·@Į@É@Ź@Ż@å@ę@ē@é@ A4A5AQDRDSDTDcDsD|D}D~D‡D™D§DØD©DŖD«D¬D­D½DĆDÉDŠDŃDśDEE E ECEIEOEUEVE†EŒE’E˜E™EĮEČEŠEŃEŅEÓEŌEÕEÖE×EłE’EF F F!F'F-F3F4FIFOFUF[F\F„FŠFF–F—FĒFĶFÓFŁFŚFG GGGGGGG5GcGNHOHXHdHsHH€H‰HH–HœHH¢HØH®H“HµHæHĒHĶHŌHÕHāHčHīHōHõHI IIIIII>IhIŽJßJąJK K&K+K0K8K@KEKFKMKSKYK_KeKfKoKuK{KK‡KˆK‘K—KK£K©KŖKµK»KĮKĒKĶKĪKĻKLLLL L(L0L5L6L=LBLHLNLTLUL^LdLjLpLvLwL€L†LŒL’L˜L™L¤LŖL°L¶L¼L½L¾LųLłL’LM MMMMM&M,M3M9M@MFMMMSMZM[MdMiMpMvM}MƒMŠMM—M˜M”M§M®M“M»MĮMČMĪMÕMÖMįMēMīMōMūMNNNNNNN9NhNPP P@PAPBPCPHPPPXP]PcPdPkPpPuPzPP“P”PP¢PØP®P³PČPÉPŅP×PŻPćPčPżPžP QQQQQ2Q3Q:QPQfQ|Q’Q©QŖQ«QćQäQåQęQėQóQūQRRRRRRR$R9R:RCRIRNRSRXRmRnRwR|RR†R‹RŸR R«R°RµRŗRæRÓRŌRŪRńRSS2S9S:SY?Y@YGYNYUY\YdYeYfYhYqYrYsY|Y}Y~YŠY‹YŒYY™Zš[_]Ź]é_ģ_0€€0€€0€€0€€0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€˜0€€p˜0€€p˜0€€p0€€p˜0€€p˜0€€p˜0€€p(0€€p˜0€€˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€p˜0€€˜0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜0€€˜0€€˜0€€H˜0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€p(0€€p˜0€€p˜0€€X˜0€€p(0€€p˜0€€p(0€€p˜0€€(0€€p˜0€€p˜0€€p(0€€p˜0€€p(0€€p˜0€€p(0€€p˜0€€p˜0€€(0€€˜0€€(0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€p(0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€p˜0€€p(0€€p˜0€€p˜0€€p0€€p˜0€€p˜0€€p˜0€€˜0€€˜0€€˜˜0€€ˆ˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜0€€p˜0€€p˜0€€p0€€p˜0€€p˜0€€p˜0€€p˜0€€˜˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜˜0€€˜0€€Č˜0€€p˜0€€p0€€p˜0€€p˜0€€p˜0€€˜0€€p˜0€€˜0€€p˜0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€˜0€€p˜0€€p˜0€€p˜0€€p(0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜0€€p˜0€€˜0€€š˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜0€€p˜0€€p(0€€p˜0€€˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€p˜0€€p0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜˜0€€˜0€€p0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€p˜0€€˜0€€p˜0€€p˜0€€˜˜0€€˜0€€p˜0€€p˜0€€p˜0€€˜0€€˜˜0€€˜0€€p˜0€€0€€p˜0€€(˜0€€p0€€p˜0€€p˜0€€p0€€p˜0€€p˜0€€p0€€˜0€€˜0€€0€€p˜0€€p˜0€€˜˜0€€p©0€€©0€€©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” 0€€ų€˜0€€˜©0€€©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€ų€˜0€€p©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€ų€˜0€€ €©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ™0€€” ˜0€€˜0€€©0€€©0€€©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€ų€˜0€€p˜0€€©0€€©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€˜0€€p©0€€©0€€©0€€©0€€ ™0€€” ©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€©0€€ ™0€€” ©0€€©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ˜0€€˜˜0€€p©0€€©0€€©0€€©0€€ ™0€€” ©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€ų€˜0€€p©0€€©0€€©0€€ ™0€€” ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€˜0€€p©0€€©0€€©0€€ ™0€€” ©0€€©0€€©0€€ ©0€€©0€€ ©0€€©0€€ ©0€€ ©0€€©0€€ ©0€€©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ©0€€ ™0€€” ˜0€€˜˜0€€ €¬z0õ¬z0õ¬z0õ¬z0õ¬z0õ¬z0õ@0€€˜@0€€¬z0Ėc¤ĮG ˜@0€€ų€˜@0€€ų€˜@0€€˜˜@0€€ų€˜@0€€ų€c€uGš3›3Ģ3Ķ3ß3ą344+4,4P4Q4m4n4–4—4æ4Ą4å4ę4Ż6Ž67787z7{7Ą7Į7Ü7Ż788K8L8„8…8ŗ8»8ö8÷899>9?9l9m9—9˜9Ą9Į9›;œ;×;Ų;1<2<]<^<<<·<ø<ä<å<==6=7=T=U=w=x=™=š=ŗ=»=ß=ą=>>(>)>M>N>u>v>”>•>³>“>Õ>Ö>(@)@+@,@.@/@G@H@Y@Z@h@i@ˆ@‰@ž@Ÿ@¶@·@É@Ź@å@ę@RDSD|D}D§DØD¬D­DŠDŃD E EUEVE˜E™EŃEŅEÖE×E F F3F4F[F\F–F—FŁFŚFGGNHOHH€HœHH“HµHŌHÕHōHõHIIßJąJK KEKFKeKfK‡KˆK©KŖKĶKĪKLL5L6LTLULvLwL˜L™L¼L½LųLłLMMZM[M—M˜MÕMÖMNNP PAPBPcPdP“P”PČPÉPżPžP2Q3Q©QŖQäQåQRR9R:RmRnRŸR RÓRŌR9S:S‹TŒTÅTĘTīTļTUUJUKUxUyUŸU U¤W„WXX3X4XPXQXwXxX”X•X¹XŗXÖX×XüXżXYY>Y?YdYeYģ_¬:0®:0ŌĖTkĢ ¬:0ŌĖŒkĢ ļ ¬:0ŌĖÄkĢp!ļ ¬:0ŌĖükĢØŚ  ¬:0 ŌĖ4lĢ``¬`Č`ä``8` ¬:0 ŌĖpCõ`l`ˆ`¤`Ą`Ü`ų` ¬:0 ŌĖxEõ`d`€`œ`¢`$¢`ģ”` ¬:0ŌĖäGõ`@`\`x`”`°`Ģ` ¬:0ŌĖHõ`TA`pA`ŒA`ØA`ÄA`ąA` ¬:0ŌĖ Kõ`ÜB`ųB`C`0C`LC`„C` ¬:0ŌĖDKõ`€D`HD`œD`øD`ŌD`šD` ¬:0ŌĖ|Kõ`°F`\F`ĢF`čF` G`8?z?«?×?Ż?@A@l@œ@»@ö@A*AJAmA—A¶AäA×C1D]DD·DäDE6ETEwE™EŗEßEF(FMFuF”F³FÕF(H.HYHˆH¶HåHRL~L¬LŠL MUM˜MŃMÖM N3N[N–NŁNONP€P“PŌPōPQßRESeS‡S©SĶST5TTTvT˜T¼TųTU3UZUpU—U®UÕUīUVhVXXzXČXŅXżX Y:YćYZZ:ZÓZ[<[Å\ī\]J]x]Ÿ]¤_`3`P`w`”`¹`Ö`ü`a>ada~aėgµ·ø¹ŗ½æĮĀÄÅĘĒČÉŹĖĢĶĪĻŠŃŅÓŌÕÖ×ŲŁŚŪÜŻŽßąįāäåęēčéźėģķīļšńņóōõö÷ųłśūüżž’     !"#$&'()*+,-./02345678:;<=>?@ABCDEFGHI     ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’JKLźg¶Å蚎9±9¹9R‡u‡}‡k(’<’>’U’i’k’Ļ•ć•å•$—G—O—é˜ ™™Q¼t¼|¼÷ĮĀ"ĀYČ|ȄČMŅpŅxŅ&ŌIŌQŌß<ßDßį#į+į<ē_ēgē'ķJķRķ&ųIųRų”üµü·ü%ż9ż;żJ2`2b2ļ455Ź9ą9ā9ß>õ>÷>ļ@AAG1G3G$I:Ié@GIN=S¢Uģ_ c25ć9ų>A4G=I8N\SĀUģ_’’Į#š¤OĀ# øHĆ#DøHÄ#ü 99BKģ_AJQQģ_=*€urn:schemas-microsoft-com:office:smarttags €PlaceType€=*€urn:schemas-microsoft-com:office:smarttags €PlaceName€9*€urn:schemas-microsoft-com:office:smarttags€place€ af¹īļņóö÷·9ø9ŗ9»9Ą9Į9{‡|‡~‡‡ƒ‡„‡‚ƒ†‡?’@’C’D’l’m’o’p’ę•ē•é•ź•M—N—P—R—T—U—™™™™™™z¼{¼}¼€¼‚¼ƒ¼ Ā!Ā#Ā$Ā,Ā-Ā‚ČƒČ…Č†ČŽČČvŅwŅyŅ|ŅŅ€ŅOŌPŌRŌUŌXŌYŌBßCßEßFßJßKß)į*į,į-į1į2įeēfēhēkēoēpēPķQķSķUķ[ķ\ķOųPųSųUųZų[ųøü¹ü¾üæü<ż>żBżCżEżFżJ2c2ļ45Ź9ć9ß>ų>ļ@AG4G$I=IN8NP’P³PĘPčPüPQ1Q:QOQPQeQfQ{Q|Q‘Q’Q„Q$R8RXRlR‹RžRæRŅRŪRšRńRSSSS1SCS\SØUĀUgYhYpYsY{Y~Y‰Yģ_ao¹ gYhY~Y‰Yģ_3aaÅńŽ9ŗ9R‡~‡k‚(’?’U’l’Ļ•ę•$—P—阙Q¼}¼÷Į#ĀYȅČMŅyŅ&ŌRŌßEßį,į<ēhē'ķSķ&ųSų”üøü%ż<żJ2c2ļ45Ź9ć9ß>ų>ļ@AG4G$I=IN8NP’P³PĘPčPüPQ1Q:QOQPQeQfQ{Q|Q‘Q’Q„Q$R8RXRlR‹RžRæRŅRŪRšRńRSSSS1SCS\SØUĀUgYhYpYsY{Y~Y‰Yģ_afhYŒYģ_’’ Network SupervisorAmy Beim David O. Beim David O. Beim David BeimåŲ5…ÄC2D2š3›3œ3¢3µ3Ģ3Ķ3Ü3Ż3Ž3ß3ą3š3÷3ż34444%4+4,4>4D4J4P4Q4j4k4l4m4n4ƒ4Š44–4—4¬4³4¹4æ4Ą4Ņ4Ł4ß4å4ę4ē4é4Ż6Ž6ß67$77787C7L7U7[7n7t7z7{7•7œ7¤7«7²7¹7Ą7Į7Ö7×7Ų7Ł7Ś7Ū7Ü7Ż7ļ7ō7ū788 888,82878<8A8F8K8L8a8f8l8r8x8~8„8…8œ8”8¦8«8°8µ8ŗ8»8Ō8Ś8ß8å8ź8š8ö8÷899999999 9%9*9/94999>9?9J9O9T9Z9`9f9l9m9y9~9ƒ9ˆ99’9—9˜9¢9§9¬9±9¶9»9Ą9Į9Ā9Ä9›;œ;;Ö;×;Ų;ī;÷;<<1<2<D<K<Q<W<]<^<v<}<ƒ<‰<<<<„<«<±<·<ø<Ļ<Õ<Ś<ß<ä<å<ž<= ====2=3=4=5=6=7=?=E=J=O=T=U=`=g=l=q=w=x=„=Š==”=™=š=„=«=°=µ=ŗ=»=Ü=Ż=Ž=ß=ą=ź=ņ=ų=ž=>>>>>">(>)>6>=>C>H>M>N>[>b>i>o>u>v>|>ƒ>‰>Ž>”>•>™>”>Ø>­>³>“>½>Ä>Ź>Ļ>Õ>Ö>×>Ł>(@)@*@+@,@-@.@/@9@G@H@S@Y@Z@`@h@i@€@ˆ@‰@—@ž@Ÿ@®@¶@·@Į@É@Ź@Ż@å@ę@ē@é@RDSDTDcDsD|D}D~D‡D™D§DØD©DŖD«D¬D­D½DĆDÉDŠDŃDśDEE E ECEIEOEUEVE†EŒE’E˜E™EĮEČEŠEŃEŅEÓEŌEÕEÖE×EłE’EF F F!F'F-F3F4FIFOFUF[F\F„FŠFF–F—FĒFĶFÓFŁFŚFG GGGGGGNHOHXHdHsHH€H‰HH–HœHH¢HØH®H“HµHæHĒHĶHŌHÕHāHčHīHōHõHI IIIIIIßJąJK K+K0K8K@KEKFKMKSKYK_KeKfKoKuK{KK‡KˆK‘K—KK£K©KŖKµK»KĮKĒKĶKĪKĻKLLL L(L0L5L6L=LBLHLNLTLUL^LdLjLpLvLwL€L†LŒL’L˜L™L¤LŖL°L¶L¼L½L¾LųLłLM MMMMM&M3M@MMMZM[MdMpM}MŠM—M˜M”M®M»MČMÕMÖMįMīMūMNNNNNP PAPBPCPHPPPXP]PcPdPkPpPuPzPP“P”PP¢PØP®P³PČPÉPŅP×PŻPćPčPżPžP QQQQQ2Q3Q:QPQfQ|Q’Q©QŖQ«QäQåQęQėQóQūQRRRRRRR$R9R:RCRIRNRSRXRmRnRwR|RR†R‹RŸR R«R°RµRŗRæRÓRŌRŪRńRSS2S9S:SY?Y@YGYNYUY\YdYeYgYhYŒYYé_ģ_Ż’@HP LaserJet IIILPT1:HPPCL5MSHP LaserJet IIIHP LaserJet III,@MSUDHP LaserJet III d HP LaserJet III,@MSUDHP LaserJet III d €aaš’`aaė_@@’’Unknown’’’’’’’’’’’’G‡:’Times New Roman5€Symbol3& ‡:’Arial"ŠhCZV&3RiFFZV&'3Ū%°r3Ū%°rÄ„„Ąxx€4øXøX3ƒ„’ßHš’?ä’’’’’’’’’’’’’’’’’’’’’Ų5…’’%The Determinants of Bank Loan PricingNetwork Supervisor David Beim