• No se han encontrado resultados

II. MARCO TEÓRICO

II.1 CUESTIONES DE PARTIDA

II.1.4 Las aplicaciones

Using a 12-page questionnaire, a total of 30 broad questions were asked, which generated 166 variables in all (see annex A for the full questionnaire). The first batch of questions asked, which cut across sections A and B, investigates the characteristics of borrowers and lenders respectively that influence the bank's willingness to lend to SMEs. The inclusion of questions on borrower and lender characteristics helped to explain the perception of bankers on the demand-side and supply-side factors that drive or constrain their involvement with SMEs in Nigeria’s post-consolidated and post-crisis period, thus providing suitable answers to proposition 1 of the study. The sort of questions asked in these sections have been included by previous survey studies (e.g. Owualah, 1988; Fletcher, 1995; Cole et al., 2004; Bruns and Fletcher, 2008, etc) and seek to investigate the microstructure of small business lending by banks in terms of the relative influence of financial and non-financial characteristics of borrowers (e.g. loan purpose, loan security, firm size, financial performance, sector of operation, credit rating, strength of previous and existing bank-borrower relationships, owners’ business experience, professional training, personal guarantee, etc), as well as their compatibility with the incentives and environments facing banks in the loan approval processes. In addition, some recent studies (e.g. Ayyagari et al., 2008; Calice, et al., 2012; Berg and Fuchs, 2013) have found that SME lending in sub-Saharan Africa is also largely driven by macroeconomic factors, the degree of competition, the information environment, the legal and contractual environment and government regulatory requirements, which were also captured in the current study. Section B also included questions on changes in lending policies and risk appetite since the global financial crisis of 2008-09. The idea of including questions on the financial crisis is to see whether the risk preferences of banks have changed as a result of the crisis and the implications these may have on SME lending.

                                                                                                                         

The second batch of questions (section C) examines the determinants of SME credit terms, which helped to provide answers to proposition 2 on the risk factors influencing loan contract determination, particularly the determinants of risk premium between large (prime) customers and small firms and the incidence of collateralization on SME loans. Some studies have shown that the financing costs or risk premium for lines of credit on small business loans are determined by a plethora of factors, such as borrowing firm’s size, loan size, firm’s age, firm’s credit rating, availability of collateral or guarantees, nature of bank-borrower relationships, etc (Cowling, 1999a; St-Pierre and Bahri, 2011). These factors have been included in the survey to ascertain the loan pricing practices of Nigerian banks with respect to SME loans. Furthermore, some other studies have also empirically examined the determinants of collateral usage and the amount of collateral charged on SME loans using micro-level data. Jimenez et al. (2006) use data from the Credit Register of Banco de Espana (CIR), while Cowling (1999b) used data from a random sample of small businesses collected from an Association of British Chambers of Commerce survey. The current study includes most of the variables tested in these empirical studies in the survey such as loan size, firm size, type of customer (new or existing), riskiness of project being financed, firm’s credit rating, owner’s credit rating, competition and business cycle factors, etc. The idea is to ascertain the extent to which banks take these factors into consideration in setting collateral requirements for SME loans, since SMEs in Nigeria face perceived risks and uncertainties that make banks reluctant to provide long-term credit to them. In addition, another question asks for the types of collateral that are most frequently accepted by the banks from SME loan customers (e.g. real estate, vehicles and business equipment, goods in stock, household goods, cash and other liquid assets, bank and personal guarantees). This question has been previously asked by Beck et al. (2008b) in a survey of 91 banks in 45 countries. Their results showed some notable differences in the frequency of collateral practices between banks in developed and developing countries.

The third batch of questions was asked across two sections (sections D and E) with the aim of providing answers to proposition 3 of the study on the significance of information acquisition by loan officers and relationship lending techniques. Section D related to how lending is organised at the bank level in terms of the centralization and decentralization of lending administration functions and the role of loan officers in loan decision-making, while section E probed into the information acquisition practices of Nigerian banks and the economic value derivable from relationship lending. Specifically, the questions on the hierarchy of loan

decision making, loan officer discretion and loan officer incentives were motivated by studies that examine the loan officer’s authority as a key determinant function of the acquisition of soft information (e.g. Benvenuti et al., 2010). The questions on the nature and quality of bank- borrower relationships, marked by the frequency of interactions between borrowers and loan officers were included to test the extent of the accumulation of soft information and the use of relationship lending techniques in loan decision-making (Cole, 1998; Boot, 2000). The questions on the economic benefits and costs of relationship lending were derived from studies that show the link between information accumulation and the value generated by banks from investing in lending relationships (Berger and Udell, 1995; Berlin and Mester, 1998; Peek, 2007; Bharath et al, 2007; Uchida et al., 2012).

The types of questions asked ranged from binary questions (i.e. Yes/No answers) to multiple- choice questions (tick only one answer), to multiple selection questions (i.e. tick all that apply) and to ranking questions (strongly disagree, disagree, agree, strongly agree). The questionnaire also included frequency-type questions (e.g. Never, Sometimes, Often, Always), as well as quantitative variables. The survey form also allowed for free-form comments for some questions to provide explanation and depth to the initial answer. All the responses, including those in the form of Likert scales were then loaded onto a statistical package (SPSS) and coded to generate the variables in variable view. In survey research, a Likert scale is an approach to response categories that measures the extent of a person’s satisfaction or agreement with a set of statements or questions. This type of response category makes it easy to quantify survey responses, simplifying data analysis. A variety of options for analyzing Likert scale data exists, including the chi square statistic, which compares respondents’ actual responses with expected answers.

The variables generated from the study included nominal and ordinal categorical variables, and continuous (scale) variables. Ordinal variables can either be string (alphanumeric) or numeric values that represent distinct categories (e.g. 1=strongly disagree, 2= disagree, 3= agree, 4= strongly agree). Values fall within discrete but ordered categories – i.e. the sequence of categories itself has meaning (Pryce, 2005). Nominal variables are data values that represent categories with no intrinsic order. The sequence of categories is arbitrary and so ordering has no meaning in and of itself (e.g. name of bank or type of branch). The complete coding framework for each of the variables can be found in annex B (at the end of the thesis).