At the most individual level, those operating within firms have a shared interest - the want to improve. Since the foundational ideas of Adam Smith and the division of labour, economists have taken an interest in attempting to understand what the firm does and how exactly it operates. The pioneering work of Coase (1937) explored the nature of the firm and the economic understanding of exactly why individuals choose to form firms to undertake business, rather than conducting market operations on an individual contractual basis. This marked the beginning of what was and has become a broad and diverse area of the literature. While the notion may have been hinted at and mentioned implicitly, the term transaction costs, coined by and developed by Williamson (1979) and Williamson (1981) made apparent the costs associated with initial contracts with suppliers and the importance of relational and ownership implications over time between buyer and seller. More generally, in the area of contract theory, theoretical applications of the principal-agent problem, in relation to the firm, have considered the design of contracts that maximise incentives on behalf of the worker, as well as considering the moral hazard that presents when information asymmetries are present (Hölmstrom, 1979; Grossman and Hart, 1983). In the advent of novel econometric techniques in the 1970s, such as stochastic fron- tier (Battese and Corra, 1977) and data envelopment analysis Charnes et al. (1978), made possible empirical studies of firm performance, namely efficiency in cost and production. While at the start of this literature, such techniques were solely used as a means to evaluate a numerical measure of efficiency and rank the observed individual units, more recent methodological advances have allowed an additional dimension of performance and efficiency studied. Using either a two or one-stage estimation method (Kumbhakar and Lovell, 2003), empirical researchers are able to use the efficiency estimates, attained from either the cost or production function, as a measure that could be explained by other explanatory covariates. This develop- ment was critical as it allowed the economist, policy maker and firms themselves, an insight into exact what attributes of the firm and the business environment either im- prove or hinder their performance. Aside from individual desires to make profit, the importance of understanding firm efficiency and its determinants has much wider implications. Given that small and medium-sized enterprise dominate the composition of firms within the private sector, particularly in developing countries; in combination with the fact that research has indicated that higher levels of GDP per capita in countries with larger numbers of SMEs (Beck et al., 2005; Ayyagari et al., 2007), there is a larger importance to understanding how we can encourage efficiency and performance. Moreover, despite the advent of new techniques, firm-level data that can be em- ployed in this framework have been relatively limited. While the applied SFA literature has been primarily concerned with the measure measurement of efficiency within utilities and agriculture, some firm or industry-level studies have emerged. That being said, those that do exists lack a expansive coverage of countries at different stages of economic and institutional development, as such, the application of policy recommendations are particularly constrained. This thesis attempts to fill a number of gaps within the existing body of literature in a number of ways. Firstly, we employ the use of a novel and comprehensive dataset. The World Bank Enterprise Survey (WBES) has an extensive coverage 139 countries, at differing stages of economics and institutional development and transition from 2006-2016. The pooled cross-sectional dataset is multifaceted, in that, it includes a range of accounting and production measures, as well as a number of typical and atypical firm characteristics. Most notable, the survey ask firm managers about their perception to a range of obstacles to firm performance, including corruption, access to finance, the informal economy and business licensing. To our knowledge, this particular insight is unique within the data available on firms at any level of coverage. Secondly, in previous studies of corruption more generally, the established mea- sures of corruption are usually at the country-level in the form of indices, namely the Corruption Perceptions Index or International Country Risk measure. In this thesis, not only do we have information about the level of corruption at the level of the firm but we also know what the perceived level of corruption that firm has experience. Similarly, with respect to financial constraints, while the availability of data measur- ing access to finance at the firm-level is more prevalent, disaggregation by severity is less so. Moreover, the characteristics of the firm included in the data set, go far beyond usual variables such as age and size. Information on tax inspections, time spent on regulatory matters, use of IT communication in client dealings and external audits are provided, that allow far more a detail analysis of the determination of firm efficiency that has also been observed in the literature thus far. In the first of three studies in this thesis, we employ a one-stage SFA approach to both calculate and subsequently attempt to understand the role of corruption and financial constraints in the variation amongst firms in productive efficiency. In doing so, we test our hypotheses. Based on a survey of the literature, we formulate two hypotheses; the severity of both corruption and constraints to finance will reduce firm efficiency. The results of our empirical analysis find that while higher levels of severity in financial constraints reduce the level of efficiency, we find evidence to the contrary for the minor levels of perceived corruption. While the former result is in agreement with the established literature, our result regarding corruption is in line with the ‘grease money’ hypothesis. In lieu of the negative effects on efficiency found in chapter two, in chapter three, we endevour to understand what determines the severity to financial constraints the firm faces, particularly different information types, soft and hard. Based on the work of Beck et al. (2006), we employ a generealised ordered probit methodology to assess the marginal effects of different information types. In contrast with Beck et al. (2006), we find evidence against using the ordered probit over the generealised model, allowing a relaxing of the parallel lines assumption. In essence we find that, despite a documented over-reliance of hard information used in the consideration of loan applications Cole et al. (2004), soft information has important implications which are otherwise disregarded in the decision-making process. For completeness, we create two scales that assess the total number of problems firm report to face, as well as their aggregate severity. Just as is the case in considering only access to finance, we find that both information types are important to both severity and number of obstacles to operation that firms must endure. Following from this, in chapter four we employ two novel approaches to construct a index measures of obstacles that firms face. By using these approachs, namely generalised structural equation modelling (GSEM), empirical Bayes prediction and polychoric principal component analysis, we construct two measure of firm obsta- cles that are manifested within the 15 perceived obstacles measures that the survey measures, given the oversimplification that simpler arithmetic measures embody. In the construction of these measures, we find the perceived obstacles with the largest weighting are corruption, courts and business licensing, in line with some theoretical predictions made in the corruption and bribery literature (Guriev, 2004). Once con- structed, we use these measures, as well as a number of other firm characteristic to assess the effect on firm performance, namely firm sales. Initially, we find that the polychoric principal component measure to be statistically preferred to the empirical bayes prediction. For the sample as a whole, we find a negative significant negative linear relation between the firm’s placement on the scale and the level of firm sales. This persists also above the 40th percentile of firm sales, as well as in particular regions, sectors and firm sizes. Finally, chapter five will summarise the thesis.