The impact of patient capital on job quality, investments and firm performance: Cross-country evidence on long-term finance

Sommer, Christoph Working Paper The impact of patient capital on job quality, investments and firm performance: Cross-country evidence on longterm finance Discussion Paper, No. 6/2021 Provided in Cooperation with: German Institute of Development and Sustainability (IDOS), Bonn Suggested Citation: Sommer, Christoph (2021) : The impact of patient capital on job quality, investments and firm performance: Cross-country evidence on long-term finance, Discussion Paper, No. 6/2021, ISBN 978-3-96021-143-3, Deutsches Institut für Entwicklungspolitik (DIE), Bonn, https://doi.org/10.23661/dp6.2021 This Version is available at: https://hdl.handle.net/10419/228839 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Discussion Paper 6/2021 The Impact of Patient Capital on Job Quality, Investments and Firm Performance Cross-Country Evidence on Long-Term Finance Christoph Sommer The impact of patient capital on job quality, investments and firm performance Cross-country evidence on long-term finance Christoph Sommer Bonn 2021 Discussion Paper / Deutsches Institut für Entwicklungspolitik ISSN (Print) 1860-0441 ISSN (Online) 2512-8698 Except as otherwise noted this publication is licensed under Creative Commons Attribution (CC BY 4.0). You are free to copy, communicate and adapt this work, as long as you attribute the German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) and the authors. Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.d-nb.de abrufbar. The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available on the Internet at http://dnb.d-nb.de. ISBN 978-3-96021-143-3 (printed edition) DOI:10.23661/dp6.2021 Christoph Sommer is a researcher in the cluster “World Economy and Development Financing” of the research programme “Transformation of Economic and Social Systems” at the German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE). His work focuses on SME finance and financial system development. Email: [email protected] Published with financial support from the Federal Ministry for Economic Cooperation and Development (BMZ) © Deutsches Institut für Entwicklungspolitik gGmbH Tulpenfeld 6, 53113 Bonn +49 (0)228 94927-0 +49 (0)228 94927-130 Email: [email protected] http://www.die-gdi.de Preface This Discussion Paper is part of DIE’s research project “Preconditions for Sustainable Development: Social Cohesion in Africa”. Social cohesion – or social solidarity – within societies is a key success factor for sustainable development in Africa. However, social cohesion is also particularly under pressure in African societies and other world regions. The DIE team aims at identifying patterns of social cohesion in Africa, analyses factors that influence the degree of social cohesion (or its absence) and identifies domestic and international policies that contribute to the creation and consolidation of social cohesion. The team addresses five issue areas: 1) Measuring social cohesion in African societies across countries; 2) Effects of tax systems and social policy on strengthening social cohesion in Africa; 3) Interdependence of financial systems design (small and medium-sized enterprises) and social cohesion; 4) Relevance of values, democracy and political institutions for social cohesion; 5) Influence of external peacebuilding, political institutions and individual attitudes on societal peace and social cohesion. This research is funded by the German Federal Ministry for Economic Cooperation and Development (BMZ). We hope that DIE research will help to better understand the drivers of social cohesion and to formulate policies that contribute to cohesive societies worldwide. Bonn, December 2020 Julia Leininger Programme Director “Transformation of political (dis-)order” and co-lead of the research project “Social cohesion in Africa”. Armin von Schiller Co-lead of the research project “Social cohesion in Africa” and senior researcher in the programme “Transformation of political (dis-)order”. Acknowledgements This Discussion Paper has been written as part of the research project “Social Cohesion in Africa” which is supported by funding from the German Ministry for Economic Cooperation and Development (BMZ). I want to express my gratitude to DIE colleagues Armin von Schiller, Jakob Schwab and Christoph Strupat for their comments. Any remaining inaccuracies are, of course, the responsibility of the author alone. Bonn, December 2020 Christoph Sommer Abstract Despite its importance for development, long-term finance is particularly scarce in countries with lower income levels. This not only results in unrealised growth and employment creation at the national level and at the level of individual firms, but also undermines a broader shift towards better jobs. After all, many long-term investments comprise investments in labour that have the potential to contribute to improvements in the quality of jobs, through training to boost skill levels, the creation of more stable employment relationships, and the higher wages that result. This paper uses more than 17,000 firm-level observations from 73 mostly low-and middle-income countries between 2002 and 2009 to provide the first empirical evidence of the extent to which long-term finance affects the quality of jobs. Additionally, it looks into effects on investments and the performance of firms. The findings, based on inverse probability weighted regression adjustment, indicate that firms with long-term finance exhibit a share of permanent employees that is 0.9 percentage points higher, and train an additional 2.4 per cent of their production workers. The probability that firms invest in fixed assets or in innovations in their production process both increase by more than 5.5 percentage points, while employment and sales growth rises as well. The fact that the positive effects on job quality mostly disappear when defining long-term finance as loans with a maturity of more than one year instead of more than two years, underlines the importance of longer loan maturities for better jobs. Despite presenting favourable theoretical and descriptive arguments, it cannot be ruled out completely that unobservable variables affect the estimation of effect sizes. Contents Preface Acknowledgements Abstract Abbreviations 1 Introduction 1 2 Conceptual framework 3 3 Data 6 4 Method 11 5 Results 14 5.1 Results for job quality 16 5.2 Results for investments and firm performance 18 5.3 Robustness checks 20 6 Conclusions 22 References 23 Appendix 27 Tables Table 1: Summary statistics 10 Table 2: Firm characteristics by different external finance situations 13 Table 3: Covariate balance before and after propensity score weighting (for share of permanent employment) 15 Table 4: Baseline ATE of patient capital on job quality 16 Table 5: ATE of patient capital on job quality (recent long-term borrowers) 18 Table 6: Baseline ATEs of patient capital on investments and firm performance 18 Table 7: ATE of patient capital on investments and firm performance (recent long-term borrowers) 20 Figures Figure 1: Theory of change for patient capital 4 Figure 2: Empirical cumulative distribution function of loan maturity 7 Figure 3: Share of long-term finance in corporate lending and private credit 8 Figure 4: Propensity scores by treatment status (for share of permanent employees) 16 Appendix tables Table A1: Description of variables and data sources 29 Table A2: Distribution of observations by country and year 31 Table A3: Covariate balance before and after propensity score weighting (for training) 33 Table A4: Covariate balance before and after propensity score weighting (for share of production worker trained) 33 Table A5: Covariate balance before and after propensity score weighting (for share of nonproduction worker trained) 34 Table A6: Covariate balance before and after propensity score weighting (for average wage) 34 Table A7: Covariate balance before and after propensity score weighting (for investment in fixed assets) 35 Table A8: Covariate balance before and after propensity score weighting (for investment in process innovation) 35 Table A9: Covariate balance before and after propensity score weighting (for investment in product innovation) 36 Table A10: Covariate balance before and after propensity score weighting (for employment growth) 36 Table A11: Covariate balance before and after propensity score weighting (for sales growth) 37 Table A12: Baseline ATE of patient capital on job quality (additional controls) 40 Table A13: ATEs of patient capital on investments and firm performance (additional controls) 40 Table A14: Baseline ATE of patient capital on job quality (survey fixed effects) 41 Table A15: ATEs of patient capital on investments and firm performance (survey fixed effects) 41 Table A16: Baseline ATE of patient capital on job quality (LMICs subsample) 42 Table A17: ATEs of patient capital on investments and firm performance (LMICs subsample) 42 Table A18: Baseline ATE of patient capital on job quality (SME subsample) 43 Table A19: ATEs of patient capital on investments and firm performance (SME subsample) 43 Table A20: Baseline ATE of patient capital on job quality (subsample before financial crisis) 44 Table A21: ATEs of patient capital on investments and firm performance (subsample before financial crisis) 44 Table A22: ATEs of patient capital (>1 year maturity) on job quality 45 The impact of patient capital on job quality, investments and firm performance & Oehmke, 2015). This is formalised by the theoretical model of Milbradt and Oehmke (2015), which builds on the assumptions that financing terms and investment decisions are interlinked and that financing frictions increase with maturity. They show that, in equilibrium, investments are inefficiently short-term and that economic growth is lowered and shocks are amplified. Long-term loans also have the potential to improve job quality, which may subsequently improve firms’ long-term prospects. While investments in highly profitable long-term projects generally include investments in physical capital such as fixed assets and equipment, it often comprises complementing investments in labour as well. New equipment, technology adoption and R&D, for instance, require staff training and accumulation of human capital. Hence, patient capital affects training, as depicted in Figure 1. More generally, as a positive side effect of investment in labour, the quality of jobs can be expected to rise, reflected, for instance, in skill development through training, higher wages and more stable employment relations. Investments in training and human capital, as necessary complements to capital investments, incentivise firms to reduce staff turnover in order to fully reap the returns on investment and to reduce skill drain. As shown in Figure 1, this should increase the share of permanent jobs within a firm and potentially even raises wages as a means of increasing the opportunity costs of switching jobs (which further increases employment stability). Existing theoretical arguments mostly underline the importance of stable employment relations for the performance of firms. Temporary employment generally raises job instability and uncertainty inside the firm, with negative effects on investment in training, internal cooperation and workers’ motivations, which, in turn, harms long-term performance and growth (Blanchard & Landier, 2002). Findings from meta-analyses detail the pathways by which temporary contracts erode the performance of firms, as illustrated in Figure 1. Temporary workers exhibit significantly lower job satisfaction (Wilkin, 2013), which negatively influences performance (Judge, Thoresen, Bono, & Patton, 2001) and turnover (Tett & Meyer, 1993). Turnover directly depresses performance (Hancock, Allen, Bosco, McDaniel, & Pierce, 2013; Park & Shaw, 2013) and additionally triggers degradation of firm-specific skills as well as underinvestment in training so that decreasing human capital further aggravates the negative effect on the performance of firms (Crook, Todd, Combs, Woehr, & Ketchen Jr, 2011). Based on the discussed theory, I expect long-term finance to improve job quality, as firms can pursue longer-term strategies with complementary investments in a stable and skilled workforce. Investments with a longer-term horizon, such as purchase of fixed assets or innovation activities, should also rise with the availability of longer-term finance. The effect on the performance of firms is ambiguous from a theoretical perspective but, given the expectations on investments in labour, physical capital and innovation, the performance of firms should increase as well. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 5 Christoph Sommer 3 Data The data stem from World Bank’s Enterprise Surveys (ES), with additional control variables from other World Bank databases, namely the World Development Indicators (WDI), Worldwide Governance Indicators (WGI) and the Financial Development and Structure Database based on Beck, Demirgüç-Kunt, and Levine (2000), Beck, Demirgüç-Kunt, and Levine (2010) and Čihák, Demirgüç-Kunt, Feyen, and Levine (2012). The strength of the ES are that they comprise nationally representative firm-level data from numerous countries with good coverage of LMICs and all sizes of firm. Only a few firms are sampled more than once, such that the data can be rather described as repeated cross-sections than unbalanced, firm-level panel data. Firms included in the ES need to be formally registered and generally number five or more employees. Most of the firms are from the manufacturing and services sectors, while agricultural and financial sectors have been completely excluded. A standardised questionnaire allows for cross-country comparisons. In this study, the dataset based on the old standardised questionnaire for the period 2002–2005 has been combined with the dataset based on the new standardised questionnaire used from 2006 onwards. It was verified that questions and variables are compatible across the old and the new questionnaire. This was cross-validated via ES panel datasets bridging the two periods of the old and new questionnaires, and by checking that variables are actually the same, correspond in their respective values and can thus be fused.3 The only exception, where corresponding variables could not be found, is the share of production/nonproduction workers receiving training, since the old standardised questionnaire employed different subcategories (share of skilled/unskilled workers). Otherwise, the data-cleaning process underlined the data quality both with regard to internal consistency and missing values. The key explanatory variable, patient capital, is based on the loan-maturity variable from the ES dataset. In the main analysis, all loans with a duration of more than two years are coded as patient capital. The robustness check also reports results when defining long-term as having a maturity of more than one year. The chosen two-year definition deviates from the more commonly used categorisation in balance sheets, reports and datasets based on the one-year threshold (e.g. Gutierrez et al., 2018; Leon, 2018)4 since it is better suited to address my research question. Firms with long-term finance are more likely to be empowered to pursue long-term growth strategies such as productivity-enhancing investments in capital (machinery, technology, etc.) and labour (training, human capital, etc.). Those investments generally require a planning horizon beyond two years and in most cases this implies the need for respective planning security in the form of financing with similar timelines. Loans with shorter maturities, in contrast, are likely to create pressures 3 For Albania, for instance, the panel data encompasses the years 2002, 2005, 2007, 2009 and 2013, and thus bridges the periods of the old and the new standardised questionnaire. Using the unique identifier for every observation (idstd), one can identify corresponding observations from the panel data and the old standardised or new standardised dataset respectively. This allows for a cross-validation of values and certification that variables from the old and the new standardised questionnaire measure the same thing and were fused correctly. Please note again, as indicated above, that the data hardly qualify as panel data: for Albania, only 188 out of the more than 1,000 firms have two or more observations. 4 Note that sometimes there is further differentiation between short-term finance (up to one year), mediumterm finance (1–5 years maturity) and long-term finance (more than 5 years). Interestingly, even when such categorisation is offered, the default for reversion to two categories is to lump together the two longer-term categories (i.e. the typical differentiation between up to one year maturity as short-term and one year plus as long-term). German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 6 The impact of patient capital on job quality, investments and firm performance for short-term profit maximisation, which may not allow firms to pursue such long-term growth strategies, with subsequent effects on job quality and the performance of firms. As a proxy for long-term finance, I use maturity of more than two years. About 49.8 per cent of firms in my sample have such a long-term loan, implying that slightly more than half of the firms rely on short-term finance. The distribution of loan maturities is illustrated by the empirical cumulative distribution function in Figure 2. Figure 2: Empirical cumulative distribution function of loan maturity Source: Author based on data from Enterprise Surveys Unfortunately, data on loan maturity are only available from 2002 to 2009, excluding 2008. Even though the variable was discontinued from 2010 onwards, its quality is very promising. First of all, the number of missing values is relatively small and amounts to less than 6.7 per cent over the 96 country-year couples included in this study (of course, these 6.7 per cent of observations with missing values for the key explanatory could not be included in the analysis). For comparison, another numerical variable that describes a loan characteristic and was continued in the ES, namely the value of required collateral, exhibits 8.3 per cent of missing values over the same sample. Moreover, the ES loan maturity variable is not taken at its face value, but merely used to create the patient capital dummy, which is one for firms with a loan of a duration of over two years. This dummy aligns very well with country-level data on maturities of the private credit portfolio. In Figure 3, ES data are aggregated to the country-year level as share of firms with long-term loans, and is plotted against the share of long-term finance in the private credit portfolio of the German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 7 Christoph Sommer corresponding country-year couple using the maturity data from Gutierrez et al. (2018).5 Even though their dataset is the most comprehensive on national loan maturity structures, it covers only 43 of the 96 country-year couples from my sample. For these observations, the correlation amounts to r=0.77 and most data points fall into a relatively narrow band around the dotted diagonal. Even with perfect data quality, we would not expect the points to fall unto the diagonal. After all, the share of long-term finance in corporate lending would only perfectly mirror the respective share in the wider national credit portfolio if long-term finance was distributed equally between household and corporate lending. However, the fact that the shares of long-term finance for firms do not deviate too much from the share of long-term finance in the national private credit portfolio raises confidence in the loan duration variable of the ES and the patient capital variable derived from it. Figure 3 further illustrates the share of long-term finance for several country-year couples in my sample and reveals the tendency that availability of long-term finance increases with the national income level. Figure 3: Share of long-term finance in corporate lending and private credit Source: Author based on data from Enterprise Surveys Gutierrez et al. (2018) define long-term finance as having a maturity of more than one year. Although this differs from the definition of long-term finance that I applied to the ES data, the figures align very well, as depicted in Figure 3. It is noteworthy that the fit is worse when applying the definition of long-term finance as loans with more than one-year duration to the ES data (correlation coefficient r=0.66). The respective graph is provided as Figure A1 in the appendix. This provides further support for defining only loans with a duration of more than two years as long-term finance or patient capital. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 5 8 The impact of patient capital on job quality, investments and firm performance The outcome variables also stem from the ES dataset and can be organised into the broader categories of job quality, investments and the performance of firms. It is challenging to adequately measure decent work and working condition, but this paper follows Blanas et al. (2019) and approximates job quality by indicators for the share of permanent employees, training and average wage. A higher share of permanent jobs take away the insecurity and pressures associated with temporary employment. Training contributes to skill development and reveals the firms’ willingness to foster their employees. It is measured by a dummy indicating whether the firm offered formal training in the last fiscal year as well as by one variable for the share of production workers and one for the share of nonproduction workers that received such training. Lastly, better pay is associated with better jobs. The average wage is computed from the total labour costs divided by the number of employees. In order to make it comparable across countries, it is set in relation to the national GDP per capita.6 Investments are more immediate outcomes from accessing external finance and include, first, investment in machinery, vehicles, equipment, land or buildings, which are captured by a dummy for whether the firm purchased fixed assets. Second, they include investments in innovation measured by a dummy for whether new and/or significantly improved products were introduced over the last three fiscal years and a dummy for the respective equivalent for production processes. Less immediate outcomes are the performance of firms as reflected in employment and sales growth. The growth rates are derived as annual averages from employment and sales figures in the last fiscal year and three fiscal years ago following Léon (2020). Sales were deflated with the GDP deflator from the World Development Indicators (WDI), and both growth rates were computed in a manner to avoid the regression-to-the-mean effect described by Haltiwanger, Jarmin, and Miranda (2013).7 Summary statistics are presented in Table 1 and indicate that the average share of permanent employees amounted to 88 per cent. The majority of firms (55%) offered training, of which roughly 22 per cent of production and nonproduction workers benefited. Average wage was slightly higher than GDP per capita but exhibits a lot of variance. About 70.5 per cent of firms purchased fixed assets and roughly half of the firms innovated and employment grew faster (5%) than sales (1.9%). Firm-level characteristics are also from the ES database and correspond to the controls commonly used in the literature on firms’ access to finance (e.g. Beck, Demirgüç-Kunt, & Maksimovic, 2008; Love & Martínez Pería, 2014). They encompass the size and age of firms, along with dummy variables for the manufacturing sector, exporters, foreign-and government-owned firms and firms with audited financial statements. As depicted in Table 1, the median firm has 38 employees and 14 years of age. Slightly fewer than half of the firms have patient capital, roughly two thirds belong to the manufacturing sector and fewer than a third export at least 10 per cent of their output. The majority of firms have audited financial statements and only 9 per cent are foreign-owned and 3 per cent governmentowned. Firm characteristics disaggregated by treatment and control group are presented in Table 2. The choice of country-level controls is informed by the same literature and comprises inflation and GDP per capita. For the first step in the estimation (propensity score 6 For the number of employees, temporary employees were converted into permanent, full-time equivalents. Furthermore, current GDP per capita in local currency was used from WDI since total labour costs from the ES database are also denominated in current local currency units. 7 The regression-to-the-mean effect is avoided by dividing not by the initial value (sales/employees three fiscal years ago), but by the average of the initial value and last value (sales/employees in last fiscal year). German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 9 Christoph Sommer model of having patient capital, see method section), additional variables are included: private credit relative to GDP, measures for competition in the banking sector (bank concentration, bank overhead costs, net interest margin) and for quality of contract enforcement, property rights and the courts (rule of law) as well as GDP growth. Details on the definition and sources for all the variables are provided in Table A1 in the Appendix. Table 1: Summary statistics Variable N mean sd min p50 max Outcome variables Share of permanent employees 17,057 0.881 0.210 0 1 1 Training 14,554 0.548 0.498 0 1 1 Share of production workers trained 12,733 22.338 35.811 0 0 100 Share of nonproduction workers trained 12,730 21.983 35.678 0 0 100 Average wage 9,628 1.093 0.826 0.047 0.889 3.929 Fixed asset investments 13,438 0.706 0.456 0 1 1 Product innovation 13,691 0.504 0.500 0 1 1 Process innovation 13,192 0.490 0.500 0 0 1 Performance growth 14,797 4.997 12.242 -32.099 2.899 47.619 Sales growth 11,328 1.884 16.887 -55.552 0.473 61.364 Firm characteristics Patient capital 17,057 0.498 0.500 0 0 1 Firm size (employees) 17,057 188.459 1059.911 1 38 67,600 Age 17,057 20.314 18.183 1 14 201 Manufacturing 17,057 0.673 0.469 0 1 1 Exporter 17,057 0.289 0.453 0 0 1 Foreign-owned 17,057 0.092 0.289 0 0 1 Government-owned 17,057 0.039 0.193 0 0 1 Audited financial statement 17,057 0.561 0.496 0 1 1 Country-level variables GDP per capita 17,057 8,452.11 9,847.14 225.62 5,693.27 52,276.2 Inflation 17,057 7.734 5.634 -7.594 6.498 24.193 Private credit per GDP 17,057 44.314 32.816 4.179 32.633 143.365 Bank concentration 17,057 64.607 14.459 24.740 64.942 100.000 Bank overhead costs 17,057 4.251 3.176 0.883 3.789 25.081 Net interest margin 17,057 5.183 2.324 0.911 4.526 13.782 Rule of law 17,057 2.431 0.764 1.272 2.175 4.164 GDP growth 17,057 5.616 2.773 -3.979 5.445 18.333 Source: Author based on data from Enterprise Surveys Some observations had to be removed prior to the analysis: first, country-year couples for which the World Bank databases do not provide data (missing values for country-level controls); second, observations from the ES database with missing values for outcome variables or firm characteristics (firm-level controls); third, the most extreme values for employment and sales growth as well as for average wage. The last step excluded the 1 per cent at the lower and upper end of employment and sales growth rates, as routinely done in literature. For average wage, the 10 per cent at the lower and upper end were dropped, since German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 10 The impact of patient capital on job quality, investments and firm performance the variable exhibited considerably more suspiciously low/high values that could not be rationalised by other characteristics observed. Lastly, countries with too few remaining observations (fewer than 20) and countries with only controls (or only treated) were removed before the estimation.8 The final sample comprises 17,057 firms from 73 countries for the period of 2002 to 2009. 9 (For details of how observations are distributed across country-year couples, see Table A2 in the Appendix.) The sample is slightly tilted towards lower-middleincome countries (44% of observations) and upper-middle-income countries (33%), with fewer observations for low-income (13%) and high-income countries (10%). 4 Method In order to identify causal effects of patient capital on job quality, investments in fixed assets and innovation as well as on the performance of firms, one needs to control for confounding characteristics of the firm and the country-specific political and economic context. Accurate estimation would ideally build on random assignment of patient capital to firms in order to ensure balanced characteristics between treated firms (𝑑𝑑𝑖𝑖=1, i.e. with patient capital) and untreated firms (𝑑𝑑𝑖𝑖=0, i.e. with short-term finance). In my context of observational data from ES, selection bias may occur, as observable and unobservable characteristics affect both the likelihood of receiving treatment and the outcome variables. The chosen inverse probability weighted regression adjustment (IPWRA) model identifies treatment effects in observational data by reweighting based on the propensity scores (Imbens & Wooldridge, 2009). More weight is given to observations that were unlikely to receive treatment (or respectively likely to receive treatment), but ended up in the treatment group (or respectively in the control group). As a consequence, balancing between treated and untreated observations and some quasi-random distribution of treatment and control is achieved. Even though IPWRA only balances according to observable variables, theoretical arguments and descriptive statistics suggest that unobservables may not differ too much across the two groups. Since only observable covariates can be included in the estimation of the propensity scores, unobservables may still introduce endogeneity problems when estimating the effects of long-term finance (e.g. Caprio & Demirguc-Kunt, 1998; Léon, 2020). This means that for unbiased estimation, unobservable variables need to be correlated with the observables such that the balancing properties extend to the unobservables as well (or need to be balanced already). By definition, the conditions of unobservable variables cannot be tested. However, there are theoretical and descriptive arguments indicating that treatment and control may not differ too much with regard to unobservables. One commonly discussed unobserved confounder in the context of (long-term) finance and the performance of firms is the quality of firms’ management (World Bank, 2015). The theoretical literature suggests that the quality of the firm – which includes the unobservable quality of the management – 8 Results are very similar when countries with too few observations are not removed. Note that for some outcome variables, the estimation strategy requires dropping at least some of these cases (e.g. due to too low propensity scores). This also motivated the exclusion of countries with too few observations. 9 Note that the sample size varies in accordance with the data availability of the outcome variable of interest. The biggest sample in the baseline analysis is realised for the share of permanent employees with 17,057 firms from 73 countries, the smallest sample of 9,228 firms for the share of nonproduction workers that received formal training. The smallest country coverage with 53 countries materialises for average wage. The number of included firms and countries are presented in the respective output tables. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 11 Christoph Sommer does not necessarily allow for conclusions on the respective loan maturities. Of course, firms need to surpass a certain quality threshold to access external finance and the threshold is probably higher for long-term loans. Yet the pool of applicants for short-and long-term finance might not be too different, according to economic theory. The decision whether to borrow short-or long-term depends on the firms’ needs arising from maturity matching and rollover risk (e.g. Graham & Harvey, 2001). The quality of the management could be related to the demand for long-term finance, since better managers may see and create more longterm investment opportunities and would thus – if they should opt to match maturities – demand more long-term finance. However, it is further argued that firms with good growth potential – which is probably associated with good-quality management – are best-suited to short-term borrowing. The reason for this is that high-growth firms will benefit less from their investment if they have to share returns with their lenders for a longer time (Myers, 1977); firms with good growth potential will also benefit from short-term loans in the context of asymmetric information even for long-term investments, as positive news on their growth will lead to better financing terms when rolling over credits Diamond, 1991). Taken together, the theoretical arguments support the notion that firms applying for patient finance are not necessarily of much better (observed and unobserved) quality than firms applying for short-term finance. Hence, even though financial institutions probably cherry-pick goodquality firms for long-term loans, there are also high-quality firms among the applicants for short-term finance that will subsequently receive loans with short maturities. This notion is underscored by descriptive statistics in Table 2. Panel A compares firms with short-term finance to firms with patient capital, along with some observable key characteristics such as the age and size of the firms, the experience of managers and the like. The maturity groups are not that different. When comparing the means via t-tests, there is no statistically significant difference for half of the variables, one is marginally significant (age) and three exhibit statistically significant differences on the 1 per cent and 5 per cent levels (government-owned, audited financial statements and experience of manager). When using a measure that is not influenced by the sample size, the standardised mean difference (Austin, 2011), only audited financial statements is found to be significantly different, while all other variables stay below the value of 0.1 commonly used in literature for significant differences. Furthermore, it is noteworthy that the minor difference in experience of manager is in favour of firms with short-term finance. In short, firms with short-and longterm difference are not too different with regard to observables. In line with the theoretical arguments, however, stark differences emerge when comparing firms with loans to the group of firms without loans, as depicted in Panel B of Table 2: except for foreign-owned, all differences are highly significant in the t-test and substantial. (Also, for the standardised mean difference all except for size and foreign-owned have a value above 0.1.) Assuming that differences exhibit similar patterns for unobservables, the descriptive statistics suggest that the endogeneity problem is much stronger when estimating the effect of finance (i.e. comparing firms with and without loans as, for example, in Ayyagari et al. (2016)) than for estimation of the effect of long-term finance (i.e. comparing firms with long-term finance to those with short-term finance, as done here). German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 12 The impact of patient capital on job quality, investments and firm performance Table 2: Firm characteristics by different external finance situations Panel A: firms without/with patient capital (firms without patient capital do have shortterm loans) Panel B: firms without/with loan mean (without) mean (with) mean diff (t-test) Stand. mean diff mean (without) mean (with) mean diff (t-test) Stand. mean diff Size (employees) 181.342 195.620 -14.277 0.01 76.774 197.169 -120.395*** 0.03 Age 20.050 20.581 -0.531* 0.03 18.152 21.295 -3.143*** 0.19 Manufacturing 0.675 0.672 0.004 -0.01 0.551 0.615 -0.063*** 0.13 Exporter 0.286 0.291 -0.005 0.01 0.161 0.283 -0.121*** 0.30 Foreign-owned 0.093 0.091 0.002 -0.01 0.084 0.085 -0.001 0.00 Governmentowned 0.043 0.034 0.009*** -0.05 0.007 0.024 -0.017*** 0.14 Audited financial statements 0.523 0.598 -0.075*** 0.15 0.441 0.593 -0.152*** 0.31 Experience of manager 15.728 14.905 0.823*** -0.07 16.770 18.425 -1.655*** 0.14 * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Based on the outlined theoretical and descriptive considerations, balancing observables via IPWRA should suffice in this context to get good estimates for the effect sizes. First, propensity scores 𝑝𝑝𝑖𝑖𝑖𝑖𝑖𝑖 = Pr(𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖 = 1 | 𝑋𝑋𝑖𝑖, 𝑍𝑍𝑖𝑖,𝑖𝑖−1, 𝛾𝛾𝑖𝑖, 𝛾𝛾𝑖𝑖 ) for firm i in country c and year t are estimated based on the following propensity score model with probit specification: 𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖 = 𝛾𝛾𝑖𝑖 + 𝛾𝛾𝑖𝑖 + 𝛽𝛽1𝑋𝑋𝑖𝑖𝑖𝑖 + 𝛽𝛽2𝑍𝑍𝑖𝑖,𝑖𝑖−1 + 𝜐𝜐𝑖𝑖𝑖𝑖𝑖𝑖 (1) The dummy variable 𝑑𝑑𝑖𝑖𝑖𝑖𝑖𝑖 captures treatment and equals one for firms with a loan of more than two years’ maturity. The vector 𝑋𝑋𝑖𝑖𝑖𝑖 comprises firm characteristics and the vector 𝑍𝑍𝑖𝑖,𝑖𝑖−1 country characteristics. Country fixed effects (𝛾𝛾𝑖𝑖) and time fixed effects (𝛾𝛾𝑖𝑖) control for unobservable differences between countries and years respectively, which includes (close to) time-invariant effects such as institutional quality, economic shocks and similar confounders on the countryor year-level. The propensity scores 𝑝𝑝𝑖𝑖𝑖𝑖𝑖𝑖 are used to compute weights according to 𝑤𝑤𝑖𝑖 = 𝑑𝑑𝑖𝑖⁄𝑝𝑝𝑖𝑖 + (1 − 𝑑𝑑𝑖𝑖) (1 − 𝑝𝑝𝑖𝑖). The formula implies that observations are weighted by their inverse ⁄ probability. Weights of 1⁄𝑝𝑝𝑖𝑖 are used for firms with patient capital and 1⁄(1 − 𝑝𝑝𝑖𝑖) for firms without. The weights are employed in the conditional mean model: 𝑦𝑦𝑖𝑖𝑖𝑖𝑖𝑖 = 𝛾𝛾𝑖𝑖 + 𝛾𝛾𝑖𝑖 + 𝛽𝛽1𝑈𝑈𝑖𝑖𝑖𝑖 + 𝛽𝛽2𝑉𝑉𝑖𝑖,𝑖𝑖−1 + 𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖 (2) The outcome variable 𝑦𝑦𝑖𝑖𝑖𝑖𝑖𝑖 is a variable capturing either job quality (share of permanent jobs, training, average wage), investments (fixed assets, product innovation, process innovation) or the performance of firms (employment or sales growth). In case of a binary outcome variable, the probit specification has been used. The vectors of firm characteristics (𝑈𝑈𝑖𝑖𝑖𝑖 ) and country-level controls (𝑉𝑉𝑖𝑖,𝑖𝑖−1) differ slightly from the ones in the treatment model (𝑋𝑋𝑖𝑖𝑖𝑖, 𝑍𝑍𝑖𝑖,𝑖𝑖−1). An overview of the variables included in the propensity score model (1) and the German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 13 5 Christoph Sommer conditional mean model (2), along with definitions and data sources, is provided in Table A1 in the Appendix (and a brief overview given in the data section). As in the propensity score model, country-level controls are lagged since for most outcome variables (e.g. investments, expansion of output and workforce, etc.) decisions are likely to be taken with some lead time and to be therefore based on developments from the previous period. Analogous to the propensity score model, country and time fixed effects (𝛾𝛾𝑖𝑖, 𝛾𝛾𝑖𝑖) are inserted. The conditional mean model is estimated separately for the treatment and the control group using the estimated propensity scores �𝑖𝑖 = 𝑑𝑑𝑖𝑖⁄  ⁄ 𝑤𝑤 𝑝𝑝 + (1 − 𝑑𝑑𝑖𝑖) (1 − 𝑝𝑝𝑖𝑖). The average 𝑖𝑖 treatment effect (ATE) is then computed as the average difference between the predicted outcomes of the treatment and the control group. One compelling feature of the IPWRA estimates is that they are doubly robust, as derived by Wooldridge (2007). This means that misspecification of either the propensity score model or the conditional mean model still results in consistency of the ATE estimates. Consistent estimation further depends on the conditional independence (CI) assumption and the overlap assumption. The CI assumption is also known as unconfoundedness and constitutes that treatment is independent of potential outcomes 𝑦𝑦(1)𝑖𝑖𝑖𝑖𝑖𝑖 and 𝑦𝑦(0)𝑖𝑖𝑖𝑖𝑖𝑖 after controlling for observables: (𝑦𝑦(1)𝑖𝑖𝑖𝑖𝑖𝑖 , 𝑦𝑦(0)𝑖𝑖𝑖𝑖𝑖𝑖 ) ⊥ 𝑑𝑑𝑖𝑖 | 𝑋𝑋𝑖𝑖, 𝑍𝑍𝑖𝑖,𝑖𝑖−1, 𝛾𝛾𝑖𝑖, 𝛾𝛾𝑖𝑖. Stated less technically, this means that beyond the observed covariates no other (unobserved) characteristics affects both treatment and outcome. Imbens and Wooldridge (2009) emphasise that this strong assumption is quite controversial, even though it underlies every multiple regression approach. In Section 5, it is shown that the CI is met for observables, as covariates are balanced between the treatment and control group after weighting (see Table 3). The second assumption is known as overlap assumption: 0 < Pr(𝑑𝑑𝑖𝑖 = 1| 𝑋𝑋𝑖𝑖 = 𝑥𝑥) < 1, for all x. It constitutes that every observation must have a positive probability of receiving any of the two treatments 𝑑𝑑𝑖𝑖=1 and 𝑑𝑑𝑖𝑖=0. Figure 4 in Section 5 shows that this assumption holds. Results As outlined in the previous section, IPWRA addresses non-random treatment allocation by balancing the covariates. Table 3 presents the standardised differences of the means and the variance ratios for all covariates before and after reweighting. It underlines the similarity of control and treatment groups. Almost all standardised differences are moved closer to zero and almost all variance ratios closer to one. Balancing has been achieved, since none of the reweighted covariates deviates more than 0.1 from these targeted values. Since the analysis comprises ten different outcome variables, and propensity scores are estimated based on the sample of the respective outcome variable and therefore differ, only the results for the variable with the largest sample (share of permanent employees) are depicted here as an example. Balancing results are equally good for the other outcome variables, as reported in Tables A3–A11 in the Appendix. This is in support of the conditional independence assumption. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 14 The impact of patient capital on job quality, investments and firm performance of the effects or the structural impact of patient capital. Yet very similar effect sizes and significance levels emerge for LMICs (see Tables A16 and A17 in the Appendix). The only difference materialises for employment growth, where the sign changes from positive to negative. Yet this change can be rather attributed to the fact that the effect was small and insignificant before (0.098pp, p=0.696), and now is even smaller in absolute terms and highly insignificant (-0.048pp, p=0.857). In a second robustness check, large firms were dropped, since they enjoy better access to long-term finance, and patient capital may play a different role for them. The results from the main analysis mostly carry over to the subsample of SMEs both with respect to effect sizes and statistical significance, as indicated in Tables A18 and A19 in the Appendix. One negligible difference is the change in sign from positive to negative for the training dummy as it used to be small and insignificant for the whole sample (0.64pp, p=0.487), and is even smaller and highly insignificant in the subsample (-0.09pp, p=0.935). The effect on investments in fixed assets, however, changes considerably. The size of the effect almost doubles from 1.8 percentage points to 3.4 percentage points, with statistical significance increasing accordingly from p=0.070 to p=0.001. Yet, this different effect for SMEs could not be confirmed when looking at the preferred specification for flow variables. When restricting the treated firms to borrowers who only took out their long-term loan recently, the effect for the SME subsample (0.0638, p=0.003) is not much different from that of the whole sample (0.0572, p=0.000). Third and last, I find no differences for the subsample restricted to the period before the financial crisis (Tables A20 and A21 in the Appendix). Lastly, the changes for alternative definitions of patient capital were explored. When defining long-term finance as loans with a maturity of more than one year, as is often done in balance sheets and subsequently in maturity datasets (e.g. Gutierrez et al., 2018; Leon, 2018), the findings from the main analysis can only partially be replicated (see Table A22 and A23 in the Appendix). For the outcome variables on investment and firm performance, effects sizes and statistical significance show only a few differences: the effect size of sales performance is much smaller and that of investment in fixed assets much larger and also more significant (p=0.015 instead of p=0.070), whereas investment in process innovation loses its significance (p=0.123 instead of p=0.024). The results, however, change considerably when looking at the job quality variables. The effect sizes of the three training indicators are much smaller and all insignificant now. The general attenuation of effects towards zero is particularly pronounced for the share of nonproduction workers trained and for average wage. Both exhibit a negative sign now and high insignificance (p=0.938 and p=0.903 respectively). The share of permanent employees is the only variable for which the effect size is somewhat similar, although even for this variable the significance is much lower for the one-year threshold (p=0.063 instead of p=0.009). Overall, it seems that effects for investments and firm performance are not too different when the definition of patient capital is altered. However, under the more short-term definition, almost all of the positive effects of patient capital on the job quality variables no longer materialise. Findings point in the same direction when moving the definition towards the three-year threshold. Effect sizes are generally estimated less precisely due to the decreasing number of firms with such patient capital (i.e. significance decreases), but effects for job quality indicators tend to be slightly higher, while effect sizes for investment and firm performance tend to be comparable to the baseline results of the two-year threshold (see Tables A24 and A25 in the appendix). The robustness check thus provides suggestive evidence that patient capital is particularly important for moving towards better jobs. This is not too surprising, given that loans with shorter maturities provide the average firm less security within which to lay plans to build up a stable and skilled workforce. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 21 6 Christoph Sommer Conclusions From a theoretical perspective, the effect of long-term finance on the performance of firms is ambiguous. Empirical evidence from the micro and macro level favours the notion that patient capital fosters investments, productivity and growth. Using firm-level data from 73 mostly low-and middle-income countries, this study provides further empirical support thereof. More importantly, it also analyses the effects on job quality. After all, many of the long-term investments, such as R&D, technology adoption and fixed assets, require complementary investments in labour, such as human capital accumulation, staff training and the like. Consequently, patient capital allows firms to pursue more long-term growth strategies, which includes investments in a stable and skilled workforce. This may contribute to better jobs, characterised by training and skill development, higher wages and more stable employment relations. Improved quality of jobs is not only a valuable goal in itself, but more broadly available good jobs also contribute to more cohesive societies (Wietzke, 2014; World Bank, 2012). The findings indicate that patient capital has indeed a positive effect on job quality. It is associated with a significant increase in the share of permanent jobs by 0.9 percentage points. Patient capital is also positively associated with formal training: an additional 2.4 per cent of production workers receive training in firms with long-term finance. The effect on average wages is positive but not significant. That fact that the effects on job quality disappear when defining patient capital as finance with a maturity of more than one year instead of using the preferred two-year threshold underlines the importance of longer-term finance in creating good jobs. Furthermore, significant and positive effects on investments and firm performance materialise. Patient capital is associated with increasing the likelihood of firms investing in fixed assets by 5.7 percentage points and by 5.6 and 3.7 percentage points for investments in process innovation and product innovation respectively. The average annual employment growth rate tends to increase by 0.77 percentage points, while the effect on the other indicator for firm performance, average annual sales growth, is positive but insignificant. Even though presented theoretical and descriptive arguments are favourable, endogeneity problems from unobservable variables cannot be ruled out completely in the estimation of the effect sizes. The results reveal that long-term finance helps to promote both employment creation and the quality of jobs. Yet additional deliberations and trade-offs need to be considered before adopting a policy agenda committed to promoting long-term finance. First, it has to be noted that it may require additional reforms and time. Markets generally require good legal infrastructure, a stable economic and political environment and functioning banking and stock markets to provide patient capital. Development finance institutions (DFIs) can play an important role in developing markets for long-term finance, but must not repeat the failures of subsidised lending from the last millennium. Second, not all firms need longterm finance, and long-term finance is more likely to go to more transparent, larger firms. This could result in a trade-off, as described by Léon (2020), that more lending with longer maturity goes to larger firms (intensive margin) at the expense of reaching more firms, in particular smaller and younger firms, with short-term finance (extensive margin). More research is needed to better understand the role of long-term finance. This refers both to exploring its relationship to job quality more thoroughly by using panel data or other means to control for unobservable firm characteristics, and the need to shed more light on the question of how to integrate reforms for long-term finance into the broader context of financial system development. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 22 The impact of patient capital on job quality, investments and firm performance References Aghion, P., Howitt, P., & Mayer-Foulkes, D. (2005). The effect of financial development on convergence: Theory and evidence. The Quarterly Journal of Economics, 120(1), 173-222. Almeida, H., Campello, M., Laranjeira, B., & Weisbenner, S. (2012). Corporate debt maturity and the real effects of the 2007 credit crisis. Critical Finance Review, 1(1), 3-58. Austin, P. C. (2011). An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behavioral Research, 46(3), 399-424. Ayyagari, M., Demirgüç-Kunt, A., & Maksimovic, V. (2008). How important are financing constraints? The role of finance in the business environment. The World Bank Economic Review, 22(3), 483-516. Ayyagari, M., Juarros, P., Martinez Peria, M. S., & Singh, S. (2016). Access to finance and job growth: Firmlevel evidence across developing countries (Policy Research Working Paper 7604). Washington, DC: World Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/24146 Beck, T., Demirgüç-Kunt, A., & Levine, R. (2000). A new database on the structure and development of the financial sector. The World Bank Economic Review, 14(3), 597-605. Beck, T., Demirgüç-Kunt, A., & Levine, R. (2010). Financial institutions and markets across countries and over time: The updated financial development and structure database. The World Bank Economic Review, 24(1), 77-92. Beck, T., Demirgüç-Kunt, A., & Maksimovic, V. (2008). Financing patterns around the world: Are small firms different? Journal of Financial Economics, 89(3), 467-487. Blanas, S., Seric, A., & Viegelahn, C. (2019). Job quality, FDI and institutions in Sub-Saharan Africa: Evidence from firm-level data. The European Journal of Development Research, 31(5), 1287-1317. Blanchard, O., & Landier, A. (2002). The perverse effects of partial labour market reform: Fixed-term contracts in France. The Economic Journal, 112(480), F214-F244. Booth, L., Aivazian, V., Demirguc-Kunt, A., & Maksimovic, V. (2001). Capital structures in developing countries. The Journal of Finance, 56(1), 87-130. Calomiris, C. W., & Kahn, C. M. (1991). The role of demandable debt in structuring optimal banking arrangements. American Economic Review, 81(3), 497-513. Caprio, G., & Demirguc-Kunt, A. (1998). The role of long-term finance: Theory and evidence. World Bank Research Observer, 13(2), 171-189. Chen, S., Ganum, M. P., Liu, L. Q., Martinez, M. L., & Peria, M. M. S. M. (2019). Debt maturity and the use of short-term debt: Evidence form sovereigns and firms (Departmental Paper 19/03). Washington, DC: International Monetary Fund. Retrieved from https://www.imf.org/en/Publications/DepartmentalPapers-Policy-Papers/Issues/2019/02/04/Debt-Maturity-and-the-Use-of-Short-Term-Debt-Evidenceform-Sovereigns-and-Firms-46240 Čihák, M., Demirgüç-Kunt, A., Feyen, E., & Levine, R. (2012). Benchmarking financial development around the world (Policy Research Working Paper 6175). Washington, DC: World Bank. Retrieved from https://openknowledge.worldbank.org/bitstream/handle/10986/12031/wps6175.pdf Crook, T. R., Todd, S. Y., Combs, J. G., Woehr, D. J., & Ketchen Jr, D. J. (2011). Does human capital matter? A meta-analysis of the relationship between human capital and firm performance. Journal of Applied Psychology, 96(3), 443. Demirgüç-Kunt, A., Horváth, B. L., & Huizinga, H. (2017). How does long-term finance affect economic volatility? Journal of Financial Stability, 33, 41-59. Demirgüç-Kunt, A., & Maksimovic, V. (1998). Law, finance, and firm growth. The Journal of Finance, 53(6), 2107-2137. Demirgüç-Kunt, A., Peria, M. S. M., & Tressel, T. (2020). The global financial crisis and the capital structure of firms: Was the impact more severe among SMEs and non-listed firms? Journal of Corporate Finance, 60, 101514. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 23 Christoph Sommer Diamond, D. W. (1993). Seniority and maturity of debt contracts. Journal of Financial Economics, 33(3), 341-368. Diamond, D. W. (1991). Debt maturity structure and liquidity risk. The Quarterly Journal of Economics, 106(3), 709-737. Diamond, D. W., & Rajan, R. G. (2001). Liquidity risk, liquidity creation, and financial fragility: A theory of banking. Journal of Political Economy, 109(2), 287-327. Duchin, R., Ozbas, O., & Sensoy, B. A. (2010). Costly external finance, corporate investment, and the subprime mortgage credit crisis. Journal of Financial Economics, 97(3), 418-435. Duval, R., Hong, G. H., & Timmer, Y. (2020). Financial frictions and the great productivity slowdown. The Review of Financial Studies, 33(2), 475-503. Fan, J. P., Titman, S., & Twite, G. (2012). An international comparison of capital structure and debt maturity choices. Journal of Financial and Quantitative Analysis, 47(1), 23-56. G20. (2013). Long-term investment financing for growth and development: Umbrella paper. Paper presented to the Meeting of the G20 Ministers of Finance and Central Bank Governors, February 2013, Moscow, Russia. Retrieved from http://www.g20chn.org/English/Documents/PastPresidency/201512/P020151225691284966192.pdf Graham, J. R., & Harvey, C. R. (2001). The theory and practice of corporate finance: Evidence from the field. Journal of Financial Economics, 60(2-3), 187-243. Gutierrez, E., Karmali, N., & Sourrouille, D. (2018). Bank credit allocation in Latin America and the Caribbean (Policy Research Working Paper 8571). Washington, DC: World Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/30323 Haltiwanger, J., Jarmin, R. S., & Miranda, J. (2013). Who creates jobs? Small versus large versus young. Review of Economics and Statistics, 95(2), 347-361. Hancock, J. I., Allen, D. G., Bosco, F. A., McDaniel, K. R., & Pierce, C. A. (2013). Meta-analytic review of employee turnover as a predictor of firm performance. Journal of Management, 39(3), 573-603. Hart, O., & Moore, J. (1995). Debt and seniority: An analysis of the role of hard claims in constraining management. American Economic Review, 85(3), 567-585. Imbens, G. W., & Wooldridge, J. M. (2009). Recent developments in the econometrics of program evaluation. Journal of Economic Literature, 47(1), 5-86. Jaramillo, F., & Schiantarelli, F. (2002). Access to long term debt and effects on firms' performance: Lessons from Ecuador (Research Network Working Paper R-460). Washington, DC: Inter-American Development Bank. Retrieved from https://core.ac.uk/reader/6441917 Jeanne, O. (2009). Debt maturity and the international financial architecture. American Economic Review, 99(5), 2135-2148. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review, 76(2), 323-329. Judge, T. A., Thoresen, C. J., Bono, J. E., & Patton, G. K. (2001). The job satisfaction–job performance relationship: A qualitative and quantitative review. Psychological Bulletin, 127(3), 376-407. Leon, F. (2018). The credit structure database (CREA Discussion Paper 2018-07). Luxembourg City: University of Luxembourg. Retrieved from http://158.64.76.181/bitstream/10993/35044/1/2018_07%20%20The%20Credit%20Structure%20Datab ase.pdf Léon, F. (2020). The provision of long-term credit and firm growth in developing countries. Economic Modelling, 90, 66-78. Li, K., Yue, H., & Zhao, L. (2009). Ownership, institutions, and capital structure: Evidence from China. Journal of Comparative Economics, 37(3), 471-490. German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 24 The impact of patient capital on job quality, investments and firm performance Love, I., & Martínez Pería, M. S. (2014). How bank competition affects firms' access to finance. The World Bank Economic Review, 29(3), 413-448. Martinez Peria, M. S., & Schmukler, S. L. (2017). Understanding the use of long-term finance in developing economies (Working Paper 17/96). Washington, DC: International Monetary Fund. Retrieved from https://www.imf.org/en/Publications/WP/Issues/2017/04/26/Understanding-the-Use-of-Long-termFinance-in-Developing-Economies-44855 Milbradt, K., & Oehmke, M. (2015). Maturity rationing and collective short-termism. Journal of Financial Economics, 118(3), 553-570. Myers, S. C. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5(2), 147-175. Park, T.-Y., & Shaw, J. D. (2013). Turnover rates and organizational performance: A meta-analysis. Journal of Applied Psychology, 98(2), 268-309. Rajan, R. G. (1992). Insiders and outsiders: The choice between informed and arm's-length debt. The Journal of Finance, 47(4), 1367-1400. Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The Journal of Finance, 50(5), 1421-1460. Schiantarelli, F., & Sembenelli, A. (1997). The maturity structure of debt: Determinants and effects on firms' performance, evidence from the United Kingdom and Italy (Policy Research Working Paper 1699). Washington, DC: World Bank. Retrieved from https://elibrary.worldbank.org/doi/abs/10.1596/18139450-1699 Schiantarelli, F., & Srivastava, V. (1997). Debt maturity and firm performance: A panel study of Indian companies (Policy Research Working Paper 1724). Washington, DC: World Bank. Retrieved from http://documents.worldbank.org/curated/en/842171468751139021/Debt-maturity-and-firmperformance-a-panel-study-of-Indian-companies Tasic, N., & Valev, N. (2008). The maturity structure of bank credit: Determinants and effects on economic growth (Andrew Young School of Policy Studies Research Paper Series 08-12, Issue. Tett, R. P., & Meyer, J. P. (1993). Job satisfaction, organizational commitment, turnover intention, and turnover: Path analyses based on meta-analytic findings. Personnel Psychology, 46(2), 259-293. Vermoesen, V., Deloof, M., & Laveren, E. (2013). Long-term debt maturity and financing constraints of SMEs during the global financial crisis. Small Business Economics, 41(2), 433-448. Wietzke, F.-B. (2014). Pathways from jobs to social cohesion. The World Bank Research Observer, 30(1), 95-123. Wilkin, C. L. (2013). I can't get no job satisfaction: Meta-analysis comparing permanent and contingent workers. Journal of Organizational Behavior, 34(1), 47-64. Wooldridge, J. M. (2007). Inverse probability weighted estimation for general missing data problems. Journal of Econometrics, 141(2), 1281-1301. World Bank. (2012). World development report 2013: Jobs. Washington, DC: World Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/11843 World Bank. (2015). Global financial development report 2015/2016: Long-term finance. Washington, DC: World Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/22543 German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 25 Appendix Christoph Sommer Comparing share of long-term finance in ES data and national maturity data Figure A1: Share of long-term finance (LTF) in corporate lending (>1 year maturity) and in private credit Source: Author German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 28 The impact of patient capital on job quality, investments and firm performance Overview of included variables Table A1: Description of variables and data sources Variable Description and data source Patient capital Dummy variable equal to one if firm has a loan with more than two years of maturity; from World Bank Enterprise Surveys (ES) Outcome variables Share of permanent employees Number of permanent, full-time employees relative to firm size (employees); from ES Training Dummy variable equal to one if employees received formal training; from ES Share of production workers trained Share of production workers that received formal training; from ES Share of nonproduction workers trained Share of nonproduction workers that received formal training; from ES Average wage Average wage, i.e. total labour costs divided by firm size (employees); from ES Investment in fixed assets Dummy variable equal to one if firm has purchased fixed assets in the last fiscal year; from ES Product innovation Dummy variable equal to one if firm has introduced a new product over the last three years; from ES Process innovation Dummy variable equal to one if firm has introduced a new or significantly improved process over the last three years; from ES Employment growth Average annual growth rate of permanent and full-time employees over the last three fiscal years; from ES Sales growth Average annual growth rate of total sales over the last three fiscal years (deflated by the GDP deflator); from ES Firm characteristics Firm size (employees) Number of full-time employees (temporary, full-time employees are converted into permanent, full-time equivalents using the average length of temporary, full-time employment); from ES Firm age Age of firm (in years); from ES Manufacturing Dummy variable equal to one if firm is in the manufacturing sector;10 from ES Exporters Dummy variable equal to one if at least 10% of firm’s output are exported (directly or indirectly); from ES Foreign-owned Dummy variable equal to one if firm is owned to 50% or more by foreign organisations; from ES Government-owned Dummy variable equal to one if firm is owned to 50% or more by the government; from ES Audited financial statements+ Dummy variable equal to one if firm’s financial statements are checked and certified by an external auditor; from ES German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 29 Christoph Sommer Table A1 (cont.): Description of variables and data sources Country-level variables Inflation Annual growth rate of the GDP deflator; from Word Bank’s World Development Indicators (WDI) GDP per capita Gross domestic product per capita (in constant US dollars); from WDI GDP growth+ Annual growth rate of GDP at market prices of constant local currency; from WDI Private credit per GDP+ Domestic credit to the private sector as % of the GDP; from the Financial Development and Structure Dataset (FDSD) Bank concentration+ Share of bank assets held by the three largest banks; from FDSD Bank overhead costs+ Banks’ overhead costs as a share of their total assets; from FDSD Net interest margin+ Banks’ net interest revenue relative to their interest-bearing assets; from FDSD [Share of foreign banks] Number of foreign banks relative to the number of total banks; from FDSD Rule of law+ Captures, amongst other things, the quality of contract enforcement, property rights, the police, and the courts; from World Bank’s Worldwide Governance Indicators + These variables are only included in the treatment model. [.] Variables in squared brackets are only included in the robustness check. Source: Author (source of variable as listed in the right column) 10 The manufacturing dummy was constructed from the ISIC codes provided in the ES data. To reduce the amount of missings, additional information was used from a meta-variable (indicating the use of the manufacturing questionnaire). German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 30 The impact of patient capital on job quality, investments and firm performance Table A11: Covariate balance before and after propensity score weighting (for sales growth) Standardised differences Variance ratio Raw Weighted Raw Weighted Firm size (employees) -0.0189 0.0018 0.9725 1.0334 Age 0.0640 0.0068 0.8934 0.9226 Manufacturing 0.0983 0.0049 0.9116 0.9954 Exporter 0.0755 -0.0056 1.0707 0.9951 Foreign-owned 0.0033 0.0018 1.0093 1.0049 Government-owned -0.0543 -0.0030 0.7883 0.9872 Audited fin. statement 0.1279 0.0004 0.9657 0.9999 Log of GDP pc 0.1723 0.0008 0.9881 0.9988 Inflation -0.1730 0.0015 0.7435 1.0073 GDP growth -0.3770 0.0012 0.7917 0.9973 Private credit per GDP 0.1053 -0.0005 0.8031 1.0013 Bank concentration -0.0351 -0.0006 1.0169 0.9982 Bank overhead costs -0.1853 0.0004 0.6403 0.9979 Net interest margin -0.1993 0.0009 0.7985 0.9989 Rule of law 0.2848 -0.0007 1.1598 1.0004 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 37 Christoph Sommer Overlap plots for the other nine outcome variables Figure A2: Propensity scores by treatment status (for job quality indicators) Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 38 The impact of patient capital on job quality, investments and firm performance Figure A3: Propensity scores by treatment status (for investment and firm performance indicators) Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 39 Christoph Sommer Robustness check: ATE for additional controls Table A12: Baseline ATE of patient capital on job quality (additional controls) Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00860** 0.00716 0.0178*** 0.00632 0.0284 (0.00386) (0.00936) (0.00664) (0.00623) (0.0269) Observations 16,769 14,232 10,464 9,018 9,400 Countries 71 69 68 68 52 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A13: ATEs of patient capital on investments and firm performance (additional controls) Investments Firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0177* 0.0266** 0.0125 0.0825 -0.262 (0.00987) (0.0112) (0.0105) (0.251) (0.371) Observations 13,223 12,870 13,369 14,529 11,083 Countries 64 63 64 69 64 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 40 The impact of patient capital on job quality, investments and firm performance Robustness check: ATE for survey fixed effects Table A14: Baseline ATE of patient capital on job quality (survey fixed effects) Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00911** 0.00601 0.0170*** 0.00600 0.0284 (0.00383) (0.00922) (0.00653) (0.00612) (0.0264) Observations 17,057 14,520 10,737 9,228 9,591 Countries 73 71 70 70 53 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A15: ATEs of patient capital on investments and firm performance (survey fixed effects) Investments Firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0175* 0.0234** 0.0144 0.105 -0.258 (0.00976) (0.0112) (0.0106) (0.253) (0.368) Observations 13,422 13,158 13,657 14,797 11,328 Countries 66 65 66 71 66 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 41 Christoph Sommer Robustness check: ATE for LMICs subsample Table A16: Baseline ATE of patient capital on job quality (LMICs subsample) Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00835** 0.00762 0.0177** 0.00593 0.0270 (0.00409) (0.0102) (0.00698) (0.00668) (0.0266) Observations 15,350 12,909 9,930 8,458 9,468 Countries 66 64 63 63 52 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A17: ATEs of patient capital on investments and firm performance (LMICs subsample) investments firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0172* 0.0227* 0.0169 -0.0482 -0.269 (0.00985) (0.0120) (0.0114) (0.268) (0.371) Observations 13,300 11,456 11,950 13,111 11,203 Countries 65 58 59 64 65 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 42 The impact of patient capital on job quality, investments and firm performance Robustness check: ATE for SME subsample Table A18: Baseline ATE of patient capital on job quality (SME subsample) Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.0113** -0.000894 0.0135** 0.00139 0.0322 (0.00484) (0.0110) (0.00661) (0.00600) (0.0276) Observations 12,082 10,082 7,931 7,052 6,974 Countries 73 70 66 66 51 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A19: ATEs of patient capital on investments and firm performance (SME subsample) investments firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0335*** 0.0222* 0.0176 0.0672 -0.314 (0.0105) (0.0133) (0.0125) (0.246) (0.453) Observations 9,288 9,114 9,521 10,751 7,672 Countries 64 64 65 71 63 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 43 Christoph Sommer Robustness check: ATE for subsample before the financial crisis 2007/08 Table A20: Baseline ATE of patient capital on job quality (subsample before financial crisis) Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00926** 0.00644 0.0197*** 0.00642 0.0201 (0.00411) (0.00949) (0.00657) (0.00662) (0.0291) Observations 15,528 13,475 9,768 8,266 8,358 Countries 70 68 66 66 48 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A21: ATEs of patient capital on investments and firm performance (subsample before financial crisis) Investments Firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0179* 0.0251** 0.0144 0.155 -0.333 (0.0105) (0.0112) (0.0105) (0.273) (0.402) Observations 11,898 13,158 13,657 13,402 10,085 Countries 63 65 66 68 62 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 44 The impact of patient capital on job quality, investments and firm performance Robustness check: Alternative definition of patient capital (>1 year maturity) Table A22: ATEs of patient capital (>1 year maturity) on job quality Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00941* 0.00225 0.00842 -0.000452 -0.00339 (0.00506) (0.0114) (0.00755) (0.00576) (0.0278) Observations 17,057 14,478 10,665 9,156 9,591 Countries 73 70 68 68 53 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A23: ATEs of patient capital (>1 year maturity) on investments and firm performance Investments Firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0298** 0.0213 0.0110 0.162 -0.0600 (0.0123) (0.0140) (0.0111) (0.267) (0.425) Observations 13,422 13,158 13,615 14,797 11,286 Countries 66 65 65 71 65 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 45 Christoph Sommer Robustness check: Alternative definition of patient capital (>3 years maturity) Table A24: ATEs of patient capital (>3 years maturity) on job quality Training (1) (2) (3) (4) (5) Share of permanent employees Training Share of production workers trained Share of nonproduction workers trained Average wage ATE 0.00946 0.0163 0.0241*** 0.00876 0.0179 (0.00691) (0.0104) (0.00746) (0.00615) (0.0240) Observations 17,057 14,520 10,737 9,228 9,591 Countries 73 71 70 70 53 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys Table A25: ATEs of patient capital (>3 years maturity) on investments and firm performance Investments Firm performance (1) (2) (3) (4) (5) Fixed assets Process innovation Product innovation Employment growth Sales growth ATE 0.0120 0.0230 0.0110 -0.0528 -0.172 (0.0129) (0.0167) (0.0116) (0.269) (0.294) Observations 13,422 13,158 13,657 14,797 11,328 Countries 66 65 66 71 66 Standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 Source: Author based on data from Enterprise Surveys German Development Institute / Deutsches Institut für Entwicklungspolitik (DIE) 46