Journal number 2 ∘ George Berulava ∘
Studying the Relationship between Trust, Trade Credit and Firms Performance in Post-Soviet Countries

Abstract

The purpose of this paper is to analyze the impact of informal trust-based relations on firm’s performance in transition economies. The trade credit variable is used as a proxy of trust-based relations and the propensity score matching method is employed to establish causal link between relational governance and business performance in the study. The research is conducted using data from a large survey of firms across 28 transition economies. The results of the study suggest that informal trust-based relations represent an important way for enhancing of business performance in transition economies.

JEL Classification: L14, D23, P31, Z13

Keywords: trust, trade credit, networks, propensity score matching, business performance, transition economies

1. Introduction.

The experience of transition economies shows that one of the key issues in the process of a market transformation of a centrally-planned system is a creation of consistent and reliable institutional framework [World Bank, 2001; World Bank, 2002].  The importance of the proper institutional framework for economic development stems from its ability to shape incentive structure of economic agents, which influences their investment and innovation decisions [Johnson, McMillan and Woodruff, 2002]. According to North [North, 1990], the existing difference in economic development levels among countries can be explained by the differences in an institutional environment, which involves both formal and informal rules of governing of a market economy. The lack of such institutions results in various types of market frictions, which hamper the productive performance of firms in transition.  In particular, market frictions such as the shortage of market information about partners and improper legal system of contract enforcement have a substantial impact on the efficiency of inter-firm relations [McMillan and Woodruff, 1999]. In the absence of sound formal institutions of contract enforcement, businesses in transition employ informal relational mechanisms of governance based on trust. Though the importance of such trust-based relations for firms in transition economies is emphasized in a significant number of academic papers its effect on the economic performance of firms remains relatively unstudied. Do firms that rely on trust in dealing with their partners are better off than firms that don’t trust their partners?  Despite its importance, there is no empirical answer to this question to the moment. One of the reasons of the lack of empirical studies of this problem is the methodological difficulty related to a determination of the causal link between trust-based governance and business performance.

The present paper seeks to fill this gap by exploring the effects of trust-based relations on business performance of firms in transition economies. To overcome the methodological problem of the causality identification, propensity score matching techniques [Rosenbaum and Rubin, 1983] is employed in the paper. The results of this study are intended to improve the understanding of the consequences of trust-based relations for the business performance of firms in transition economies, and thus they extend the existing theoretical framework.

The rest of the paper is organized as follows. Section 2 examines the existing literature in the fields of research related to trust-based relationships. Based on the literature review, the research hypothesis is formulated. In section 3 we turn to a discussion of the research methodology, including empirical strategy and measures. The data set and characteristics of variables used in the study are described in section four. The fifth section provides analysis into the study results. The final remarks are presented in section 6.

2. Literature Review.

The key element of informal or relational governance is the trust [Bradach and Eccles, 1989]. The concept of trust that underlies relational contractual arrangements is based on social norms and personal relations [Lewis, 1985]. Heide and John [Heide and John, 1990] show that norms play a very important role in structuring economically efficient relationships between independent firms. They argue that supportive norms have significant economic value when specific assets need to be safeguarded. Mitigating possibility for opportunistic behavior and reducing uncertainty, trust reduces pressure toward vertical integration [Granovetter, 1985]. Macaulay [Macaulay, 1963] in his preliminary study of non-contractual relations in business found that the norms of keeping commitments impose obligations on parties to transactions at the cost of damaging personal relationships. Arrow emphasizing the role of trust as a control mechanism defines it as "…an important lubricant of a social system" [Arrow, 1974: 23]. The role of informal trust-based institutions takes on special significance for firms operating in transition economies. Such institutions allow firms to cope with the issues of high transaction costs, uncertainty and scarce information in dealing with their partners and thus facilitate smooth functioning of the economies in transition.

The performance of trust-based informal institutions in transition economies has been explored in a number of studies. Raiser, Allan and Steves [Raiser, Allan and Steves, 2004] based on the data from a large survey of firms across 26 transition countries examine the determinants of trust in the transition process. Using ‘the level of prepayment demanded by suppliers from their customers in advance of delivery’ as a proxy for trust they confirm earlier findings that trust is higher where firms have confidence in third party enforcement through the legal system. Other findings of the study can be summarized as follows: the fairness and honesty of the courts are a more important determinant of inter-firm trust as compared to the courts’ efficiency or ability to enforce decisions; networks based around personal ties – family and friends – and business associations are important determinants of the development of trust, while business networks based on enterprise insiders and government agencies are not; country-level effects are significantly more important factors of inter-firm trust than are firm-level effects.

Berulava and Lezhava [Berulava and Lezhava, 2008] using data from a sample of Georgian manufacturing enterprises find that trust along with traditional dimensions of transaction cost economics (asset specificity and uncertainty) has a significant impact on the choice of exchange governance mode. They find that trust produced by informal institutions such as networks comprised from friends and relatives as well as from business associations play important role in facilitating relationships between manufacturers and distributors in Georgia.

 McMillan and Woodruff [McMillan and Woodruff, 1999] examining trade credit issues in Vietnam find that in a weak contract enforcement environment, informal institutions serve as a substitute to a legal system. In particular, business network formed by relatives or friends, functions as important source of information, thus generating trust and promoting exchange.  Similarly, the survey of managers of privately-owned manufacturing firms in Russia, Ukraine, Slovakia and Romania provides evidence that relationship contracting works as a substitute for the courts [Johnson, McMillan and Woodruff, 1999]. The same time, the authors find that though relational contracting was the basis of the most transactions in all of the countries, the law also did matter.  The study results suggest that information from other economic agents, long period of cooperation and high switching costs support trade credit.

Summarizing, the existing research reveals that trust-based contracting can work either as a substitute or complement for legal institutions, thus reducing transaction costs and facilitating exchange between firms. However, the literature acknowledges that such type of relationships can cause some inefficiency in firm’s performance as well.  For instance, McMillan and Woodruff [McMillan and Woodruff, 1999] argue that informal relationships come with efficiency costs, since better exchange opportunities from economic agents outside of the network could be lost. Similarly, according to Johnson, McMillan and Woodruff [Johnson, McMillan and Woodruff, 1999], relational contracting along with aiding contract can bring some inefficiency. Thus, the question of interest is the net result of the trust-based relation’s effect on the business performance. Are firms better off when they are engaged in informal relations with partners or the opposite statement is true?

Despite its importance, to the moment the overall impact of informal contract relationships on the firm’s performance remains relatively unstudied in economic literature. The existing studies of trust-based relations focus mainly on exploration of its determinants and various types of governance structures, while economic consequences of such relationship received very sparse attention from academicians. This paper aims to shed light on exactly this issue by exploring the effect of trust-based relations on various indicators of firm’s performance in transition economies. In particular the main research question of the paper is as follows:

• Do firms employing trust-based relations perform better in terms of productivity, innovations, and sales than firms not relying on such institutions? 

To get the answer on this question, first of all let’s consider the ways in which trust-based relationships can improve firm’s functioning. Sako [Sako, 2002] emphasizes three mechanisms through which trust may enhance business performance. First, trust-based relationship allows for reducing of transaction costs and thus it ensures the most efficient governance structure. Second, trust stimulates investments in specific assets, which in turn guarantees future returns and productivity growth. Third, trust encourages orientation towards joint problem solving in such matters as cost reduction, innovation, management promoting thus continuous learning and enhancement. Based on the empirical study of automotive industry in Japan and USA the author shows that supplier’s trust of customers generally is associated with its better performance in terms of costs, profit margins, just-in-time (JIT) delivery and joint problem solving [Sako, 2002]. Similarly, Dyer [Dyer, 1996] based on the results of his empirical study, emphasizes asset co-specialization and lower transaction costs (which are outcomes of trust-based hybrid/alliance governance structures) among the factors that provide Japanese automotive firms with competitive advantage over their U.S counterparts. Hendley, Murrell, and Ryterman [Hendley, Murrell, and Ryterman, 1998] in their study of transactional strategies of Russian enterprises found that during transition, strategies that use trust have a critical importance as well as personal relationships. Based on the review of the results of existing studies, the research hypothesis of the paper can be formulated as follows: employing informal trust-based relations improves overall performance of firms in transition economies.

3. Research Methodology.

Empirical Strategy. In this paper, following Johnson, McMillan and Woodruff [Johnson, McMillan and Woodruff, 1999] and McMillan and Woodruff [McMillan and Woodruff, 1999] we use a trade credit as a proxy variable for trust-based governance. It is deemed that a firm will provide a trade credit to its partner only in case when the relationships between partners are trust-based. Further, we test the hypothesis on the casual link between trust-based relations and business performance by employing propensity score matching procedure (PSM) [Rosenbaum and Rubin, 1983]. The PSM techniques utilized in this study, allows us to delineate the casual effects of trade credit on business outcomes. This method imitates a controlled experiment and assumes creation of a counterfactual that is similar to the treated population by matching them on a variety of variables in order to control for observable differences. For instance, the counterfactual question of the study can be formulated as follows: “What would have happened to the firms which, in fact, did receive ‘treatment’ (in our case the firms that trust their partners through providing them trade credit), if they had not received ‘treatment’ (no trust)?” The advantage of this approach is that it facilitates identification of the direction of causality between variables of interest. This method addresses also the selection bias issues and allows for heterogeneities and non-linearities in the effects of informal relationships on firm’s performance.

Our empirical strategy implies implementation of a number of consecutive steps. At the first stage we calculate propensity scores, to account for non-randomness in which firms provide trade credit. The propensity score allows coping with the issue of selection bias by comparing groups based on observed covariates; and thus, it represents a good tool for estimation the treatment effect when treatment assignment is not random. Propensity scores are estimated using the following logit regression for the probability that a firm gives trade credit (trusts) to its partner:

 

where y_i is i firm’s choice of the mode of relationship with partners (yi=1 if a firm provides trade credit to its partner and y_i=0 otherwise); x_i is a set of observed covariates (discussed in more detail in the next section);β^' – vector of parameters to be estimated.

According to Rosenbaum and Rubin [Rosenbaum and Rubin, 1983], comparing firms with a similar probability of providing trade credit given the observables in x_i is equivalent to comparing firms with similar values of x_i.  Thus, after calculation of propensity scores, on the next step, the actual matching procedure is conducted using the “kernel” matching technique. The advantage of this approach is that it allows for maximum use of all the observations. Based on propensity scores the matching procedure implies estimation of a counterfactual for each treated observation.

Assuming that the effect of residual factors on treatment assignment net of treatment propensity is ignorable, we can calculate the expected casual effect of the treatment (providing trade credit) on the performance of firm. This effect is known as average treatment effects for the treated (ATT). The ATT measures the effect of providing trade credit on the outcome variable for those firms that actually provided trade credit compared with what would have happened if they had not relied on trust-based relations with partners (no trade credit). For individual firm, the average treatment effect on the treated can be calculated in the following way:

 ATT=E[q_i1-q_i0|y_i=1]                (2)

where - is potential output of firm i, which is exposed to treatment (firm provides trade credit); - is potential output of firm i, which represents a control group not exposed to treatment (firm doesn’t provide trade credit).

Measures. To explore the potential impact of trust-based relations (providing of trade credit) on firm’s performance, a number of outcome variablesare used in this study for which the corresponding ATT are identified. These outcome variables reflect various aspects of firm’s performance and are constructed in different ways. These variables are:

Sales Growth – dichotomous variable is coded as 1 if over the last 36 months a firm experienced increase in sales and is coded as 0 otherwise.

 Innovation – is represented by normalized factor score, which reflects innovative activities undertaken by a firm during the last 36 months. We use principal component factor analysis to construct this variable from the following four innovation variables: developing successfully a major new product line/service; upgrading an existing product line/service; creating a new joint venture with foreign partner; obtained a new quality accreditation (ISO 9000, 9002 or 14,000, AGCCP, etc) [1].

Percentage of Reinvestment – is measured as percentage of total profits reinvested in a firm.

Labor Productivity – this variable is measured as a logarithm of the ratio of sales volume (in USD) to a number of full-time employees.

The dependent variable in the logit regression – Trade Credit – is a dichotomous variable constructed from the continuous variable which reflects the percentage of firm’s sales to customers over the last 12 months that were sold on credit. The variable is coded as 1 if more than ten percent of sales were sold on credit and coded as 0 otherwise.

The choice of covariates, used in calculation of propensity scores, is based on the theoretical framework and the existing literature [Raiser, Allan and Steves, 2004; Johnson, McMillan and Woodruff, 1999; Carlin, Schaffer and Seabright, 2004]. However, this study employs only limited number of variables in order to avoid the violation of the common support assumption. These variables are assumed to influence both the decision to provide trade credit and firm’s performance.

First, following Raiser, Allan and Steves [Raiser, Allan and Steves, 2004] we include variables that reflect existing legal system and networks.

Legal System – is constructed on the basis of principal components factor analysis using five questions, each employing 6-point scale. The respondents were asked about how often they associate the following descriptions with the court system in resolving business disputes. These descriptions are: fair and impartial; honest/uncorrupted; quick; affordable; able to enforce its decisions[2].

Factor analysis was used in construction of Network variables as well[3].  Initially network variables were measured in the following way. On a 5-point scale ranging from extremely important =5 to not important =1 respondents rated the importance of the following sources of information on new customers: family and friends; former employees who now work for a potential customer or supplier; prior employment of managers by a potential customer or supplier current distributors; existing customers or suppliers; government agencies, business associations and other sources. Similar to Berulava and Lezhava [Berulava and Lezhava, 2008] study, factor solution suggests on existing of the two types of network variables:

Narrow Networks – include information from narrow group of people such as family and friends; former employees who now work for a potential customer or supplier; prior employment of managers by a potential customer or supplier current distributors; existing customers or suppliers.

Broad Networks – include information from a broader group of sources such as government agencies, business associations and other sources. Other controls employed in the study are:

Internal Funds/Working Capital – percentage of firm’s working capital financed from internal funds or retained earnings.

Internal Funds/Working Capital - percentage of firm’s new fixed investment financed from internal funds or retained earnings. Both variables serve as proxies for capital market constraints [Raiser, Allan and Steves, 2004].

Customer Change - dummy variable for whether firm has changed its major customer in last 3 years.

Payment Delay - dummy variable for whether firms have ever experienced an overdue payment.

Sales to Government - percentage of domestic sales to government.

Sales to Multinationals - percentage of domestic sales to multinational companies located in host country.

New Firm - dummy variable for whether firms are newly established entities.

Competition – measures degree of competition using the number of competitors reported by the respondent in the market for its main product. Based on the answers, three dummy variables are created: no competitors; 1-3 competitors; more than 3 competitors.

Following Carlin, Schaffer and Seabright [Carlin, Schaffer and Seabright, 2004] we incorporate three additional variables that influence decision of firm to innovate in the models that estimate the effect of trade credit on innovation decisions of firm. These variables reflect the importance for firms while they make their decisions on the developing new products or services and markets of each of the following factors: Domestic Competitors; Foreign Competitors; Customers. Industry (manufacturing/service)and country controls[4] are also used in the study.

4.Data Set.

The main source of the data for the research is the micro-level dataset from the Enterprise Surveys program (Business Environment and Enterprise Performance Survey (BEEPS) III round)[5]. The survey was conducted by the European Bank for Reconstruction and Development (EBRD) and the World Bank Group (the World Bank) for 9,655 firms in 28 countries in the European and Central Asian region in 2005. In all countries where a reliable sample frame was available, the sample was selected using stratified random sampling. Three levels of stratification were used in all countries: industry, establishment size and region. The more detailed description of the sampling methodology can be found in the Sampling Manual[6].

Table 1, presents a description of the key variables used in the study. According to the data from this table, out of 9,655 observations, fifty-four percent of firms reported improvement of their performance in terms of sales growth over the last 36 months. 

On average, almost half of the profit earned by the firms in the sample was reinvested in firms. Another important dimension of firm’s performance used in the study is innovation. According to table 1, over the period of last three years almost thirty-five percent of firms introduced a new product line, half of the sample upgraded existing product, more than twelve percent obtained a new quality accreditation ISO, and only four percent of firms opened a new plant. The average rate of labor productivity for the sample (out of 6,984 observations) is approximately thirty-six thousand of USD per employee. Trade credit as a mean of relationships with the partners is employed by a half of the firms in the sample (out of 9,595 observations).  

The evaluation of the legal system reveals that ability of court to enforce its decisions received the highest rating, while its affordability and quickness the lowest. Among the network sources of information existing customers/suppliers have the highest level of credibility. Government agencies and business associations are the least trusted sources of information on business partners. Almost half of the firms in the sample have ever experienced an overdue payment and about twenty-two percent have changed a major customer in last three years. The share of sales to government and multinationals doesn’t exceed five percent each. Around seventy percent of working capital and new fixed investments is financed from the internal funds.

 Table 1. Descriptive statistics

Variables	Mean	Standard deviation	Number of observations
Sales growth	.540	.50	9,655
New product line	.349	.48	9,655
Upgrade of existing products	.502	.50	9,655
Opening a new facility	.042	.20	9,655
Obtained a new quality accreditation ISO	.125	.33	9,655
Reinvestment of profits (percent)	49.530	40.07	7,781
Labor productivity (thousands of USD)	35.859	160.81	6,984
Trade credit	.500	.50	9,595
Court: fair/honest	2.923	1,37	8,339
Court: quick/affordable	2.760	1.23	8,418
Court: can enforce its decisions	3.363	1.52	8,665
Information about customer: Family and friends	2.507	1.38	9,461
Information about customer: Former employees/ managers	2.303	1.17	9,136
Information about customer: Existing customers or suppliers	3.414	1.30	9,369
Information about customer: Government agencies	2.200	1.35	9,242
Information about customer: Business associations	2.154	1.31	9,246
Information about customer: Trade fairs/others	2.746	1.42	9,320
Payment delay	.504	.50	9,655
Change of major customer	.218	.41	9,655
Sales to government (percent)	4.259	14.35	9,327
Sales to multinational corporations (percent)	3.925	13,72	9,327
Working capital financed from internal funds (percent)	72.270	37.39	9,430
New fixed investments financed from internal funds (percent)	70.136	39.85	6,836
New firm	.793	.41	8,806
No competitor	.060	.24	8,411
One-to-three competitors	.239	.43	8,411
More than three competitors	.701	.46	8,411
Pressure from domestic competitors: not at all important	.136	.34	9,526
Pressure from domestic competitors: slightly important	.187	.39	9,526
Pressure from domestic competitors: fairly important	.347	.48	9,526
Pressure from domestic competitors: very important	.330	.47	9,526
Pressure from foreign competitors: not at all important	.459	.50	9,212
Pressure from foreign competitors: slightly important	.182	.39	9,212
Pressure from foreign competitors: fairly important	.190	.39	9,212
Pressure from foreign competitors: very important	.169	.37	9,212
Pressure from customers: not at all important	.121	.33	9,466
Pressure from customers: slightly important	.158	.36	9,466
Pressure from customers: fairly important	.337	.47	9,466
Pressure from customers: very important	.385	.49	9,466

New firms represent approximately eighty percent of the sample. Most of the firms (70.1 %) encounter intense competition (facing with more than three competitors), while about twenty-four percent of firms have only 1-3 rivals. Only six percent of firms reported that they have no rivals. According to table 1, pressure from domestic rivals and from customers is the most important incentive of innovation for the firms in the sample. The threat from the foreign rivals seems to be less important stimulus for innovation.

5. Study Results.

Trade Credit Prediction. The dependent variable in our logit regressions is dichotomous, which reflects whether or not firm provides trade credit to its partner. Propensity scores are calculated separately for each of four outcome variables[7]. The results of estimation of the four logit regressions are presented in table 2.

Table 2. Trade Credit Logit Regression Results

Dependent variable: trade credit/no trade credit
Covariates	Models with outcome variables:
	Sales growth	Labor productivity	Percentage of reinvestment	Innovation
Legal system	-.196 (.166)	-.169 (.204)	-.135 (.182)	-.234 (.170)
Narrow networks	-.113*** (.036)	-.099** (.042)	-.087** (.040)	-.129*** (.038)
Broad networks	.155*** (.036)	.151*** (.042)	.143*** (.039)	.109*** (.037)
Internal funds/working capital	-.003**(.001)	-.003** (.001)	-.004*** (.001)	-.002** (.001)
Internal funds/new fixed investment	-.002* (.001)	-.001 (.001)	-.001 (.001)	-.002** (.001)
Payment delay	1.060*** (.075)	1.020*** (.087)	1.016*** (.081)	1.055*** (.076)
Service	-.449*** (.074)	-.462*** (.087)	-.458*** (.080)	-.407*** (.076)
New	.019 (.094)	.121 (.106)	-.009 (.104)	.010(.096)
Sales to government	-.003 (.002)	-.002 (.003)	-.002 (.003)	-.002(.002)
Sales to multinationals	.011*** (.002)	.010*** (.003)	.011*** (.002)	-
Customer change	.205** (.083)	.156 (.096)	.075 (.091)	-
Competition_1 (no competitors)	-.094 (.180)	-.050 (.202)	-.033 (.196)	-.075 (.189)
Competition_2 (1-3 competitors)	.061 (.083)	.135 (.097)	.125 (.090)	.141 (.085)
Country controls	Yes	Yes	Yes	Yes
Pressure from domestic competitors_1	-	-	-	-.166 (.148)
Pressure from domestic competitors_2	-	-	-	-.307*** (.117)
Pressure from domestic competitors_3	-	-	-	-.151(.093)
Pressure from foreign competitors_1	-	-	-	-.555*** (.115)
Pressure from foreign competitors_2	-	-	-	-.276** (.127)
Pressure from foreign competitors_3	-	-	-	-.178 (.126)
Pressure from customers_1	-	-	-	.092 (.156)
Pressure from customers_2	-	-	-	.076 (.118)
Pressure from customers_3	-	-	-	.219** (.090)
Model fit
LR chi2(df)	861.05 (38)	660.23 (37)	740.32 (38)	862.64 (46)
Prob> chi2	0.0000	0.0000	0.0000	0.0000
Pseudo R2	0.1498	0.1560	0.1504	0.1547
Number of observations	4154	3071	3557	4029
Notes: Standard errors in parentheses;  *** — significant at p < 0.01 level; ** — significant at p < 0.05 level; * — significant at p < 0.1 level.

The explanatory power of the all regression is quite satisfactory since all models are statistically significant at one percent level and pseudo R² are above 10%. As it was expected the broad networks comprised from business associations and government agencies has a positive and statistically significant (at p

The Impact of Trade Credit on Business Performance. The kernel matching procedure for estimation of average treatment effect is used to identify the impact of the trust-based relationships (trade credit) on the business performance in this study. [8] According to the table 3, in support to the main hypothesis of the study, we find that in general, trade credit improves business performance of companies. In particular, trust-based relations (trade credit) tend to increase sales of firms. The difference for treated and control groups is above six percentage points and is statistically significant at the one percent level. Trust-based relations stimulate innovative behavior as well. The effect of trade credit on firm’s innovation is statistically significant at p[9]. Firms that trust to their partners invest more in their business. The share of reinvested profits is higher by six percent for the firms that provide trade credit to partners (statistically significant at one percent level). These firms are also more productive in terms of labor productivity compared to firms that don’t rely on trust (significant at p

Table 3. Estimated Average Treatment Effect on Treated (ATT) for Trade Credit

ATT	Outcome variables
	Sales growth	Labor productivity	Percentage of reinvestment	Innovation (factor score)
Treated	.60603	3.36259	56.40054	.29577
Controls	.54070	3.15469	50.47600	.27432
Difference	.06532***	.20789***	5.92454***	.02145**
Standard Error	.01938	.04809	1.64058	.00980
T-statistic	3.37	4.32	3.61	2.19
Notes: *** — significant at p < 0.01 level; ** — significant at p < 0.05 level; * — significant at p < 0.1 level.

Sensitivity Analysis. In this study, the significant effect of trust-based relations (trade credit) on firm’s performance is found on the basis of propensity score matching procedure. However, since PSM cannot control for unobservable characteristics, the question is whether these results are robust to unobservable variables. To say distinctly, an unmeasured confounding variable may impact selection into the treatment and thus undermine the conclusions. To find how strongly ‘hidden biases’ might affect the results of the study we employ sensitivity analysis proposed by Rosenbaum [Rosenbaum, 2002]. Since the outcome variables of different nature (both dichotomous and continuous) are used in this study, we apply two alternative procedures of sensitivity analysis: Hodges-Lehmann point estimates[10][DiPrete and Gangl, 2004] for continuous variables and Mantel and Haenszel [Mantel and Haenszel, 1959][11] test statistic for the discrete one [Becker and Caliendo, 2007]. The results of sensitivity analysis presented in tables 4 and 5 show that robustness to hidden bias varies significantly across the different outcomes.

Table 4. Rosenbaum bounds sensitivity analysis: Hodges-Lehmann point estimates for variable Trade Credit

Outcome variables	Gamma*	Significance level		Hodges-Lehmann point estimate		Confidence interval (95%)
		upper bound	lower bound	upper bound	lower bound	upper bound	lower bound
Innovation (factor score)	1	.039688	.039688	.011511	.011511	-.00089	.024168
	1.1	.556352	.000129	-.000593	.023821	-.00795	.034481
	1.2	.969456	3.4e-08	-.006949	.032859	-.016587	.0474
	1.3	.999738	1.2e-12	-.014305	.044614	-.026727	.057475
	1.4	1	0	-.023525	.054505	-.032436	.065984
	1.5	1	0	-.030376	.062149	-.039059	.074346
	1.6	1	0	-.035184	.069965	-.045978	.081305
	1.7	1	0	-.041526	.077077	-.052382	.086434
	1.8	1	0	-.04762	.082586	-.058048	.092908
	1.9	1	0	-.053162	.087132	-.063492	.100064
	2.0	1	0	-.058143	.093047	-.069299	.105434
Labor productivity	1	0	0	.236014	.236014	.192472	.279169
	1.1	0	0	.19869	.273084	.155112	.316232
	1.2	4.3e-13	0	.164651	.306743	.120728	.350033
	1.3	3.8e-09	0	.133374	.337708	.088928	.381133
	1.4	3.2e-06	0	.104324	.36609	.059458	.409722
	1.5	.000413	0	.077101	.392606	.032222	.436286
	1.6	.012154	0	.051859	.41715	.006847	.461038
	1.7	.109546	0	.028206	.440235	-.017279	.484071
	1.8	.395415	0	.006112	.461703	-.040144	.505904
	1.9	.740909	0	-.014978	.482001	-.061703	.526836
	2.0	.934632	0	-.035224	.501179	-.082063	.546281
Percentage of reinvestment	1	1.1e-16	1.1e-16	5.34297	5.34297	3.77527	6.46069
	1.1	7.3e-11	0	3.92317	6.3305	2.45332	8.02064
	1.2	7.5e-07	0	2.72005	7.68218	1.45339	9.67885
	1.3	.000405	0	1.71697	9.11178	.752616	10.8417
	1.4	.022713	0	1.01281	10.45	.030805	11.9178
	1.5	.227328	0	.437799	11.2935	-1.0688	13.5442
	1.6	.664071	0	-.337541	12.4413	-2.10398	15.1517
	1.7	.936273	0	-1.32086	13.9492	-3.14728	16.0382
	1.8	.994814	0	-2.24069	15.2914	-3.91679	17.2543
	1.9	.999806	0	-3.15281	16.0447	-4.55536	19.0195
	2.0	.999996	0	-3.83573	17.1031	-5.35743	20.3552
Note: * - gamma  - log odds of differential assignment due to unobserved factors

Table 4 reports the Hodges-Lehmann point estimatesresults for continuous outcome variables: innovation (factor score); percent of reinvestment and labor productivity. These results show that the outcomes under consideration are sensitive to potential impact of unobservable variables. For reinvestment and labor productivity outcome variables, the Hodges-Lehmann point estimates encompass zero at gamma=1.5 and gamma= 1.7 respectively. These values mean that the unobserved characteristic would have to increase the odds ratio by less than 50% and 70% respectively before it would bias the estimated impact. The situation is even worse with respect to innovation variable, the treatment effect becomes insignificant at gamma=1.1. These relatively low values (less than critical value of 2) imply that the treatment effects for reinvestment, labor productivity and especially for innovation are sensitive to unobserved characteristics. Thus some caution is needed when interpreting the results based on these findings.

The results of sensitivity analysis for discrete variable - sales growth – are presented in Table 5. According to the Table 5, the average treatment effect is statistically significant even at high levels of gamma. This means that the average treatment effect estimated for this output variable is insensitive and robust to the presence of hidden bias.

Table 5. Mantel-Haenszel bounds sensitivity analysisfor variable Trade Credit

Outcome variables	Gamma*	Mantel-Haenszel statistic		Significance level
		overestimation of treatment effect	underestimation of treatment effect	overestimation of treatment effect	underestimation of treatment effect
Sales growth	1	2.59199	2.59199	.004771	.004771
	2	8.32586	13.6621	0	0
	3	14.8527	20.3319	0	0
	4	19.5811	25.2127	0	0
	5	23.3252	29.1111	0	0
	6	26.4452	32.3851	0	0
	7	29.1326	35.2259	0	0
Note: * - gamma  - odds of differential assignment due to unobserved factors

Summarizing, the sensitivity analysis of the impact of trade credit on firm’s performance variables shows mixed results. While some output variables are sensitive to hidden bias the other are quite robust with respect to potential impact of unobserved characteristics. However, one should realize that sensitivity analysis doesn’t reveal the existence of hidden biases per se; rather it indicates how the treatment effect can be influenced by these biases.

6. Conclusions.

The purpose of this paper was to analyze the impact of trust-based relations on firm’s performance in transition economies. We use trade credit as a proxy of trust-based relations in this study. In particular, the question we seek to address in this study was, “does trade credit to customers improve business performance of firms that provide it?” The answer to this question may have important implication for the development of best business relation practices for the firms in transition economies. However, an empirical test of this issue has not been implemented to the moment because of the complications involved in establishing of a causal link between trust-based relations and business performance. The main contribution of this study is that it provides new empirical insights into the casual link between trust-based relations and business performance of firms in transition economies. In particular, we address this problem by using propensity score matching method to establish counterfactuals for firms that provide trade credit to their customers, and matching these companies with similar firms that don’t trust their customers based on characteristics that affect both the probability of choice for providing trade credit and business performance outcomes. Specifically, we employed covariates that reflect trust of economic agents to the legal system as well as to information provided by networks from friend, relatives, colleagues, partners, business associations and government agencies; degree of competition and pressure on the firm to innovate from customers, domestic and foreign competitors; variables that reflect experience of the firms in dealing with partners and a couple of financial indicators; industry and country controls. The study was conducted using data from a large survey of firms across 28 transition economies.

The results of the study suggest that informal trust-based institutions of contract governance represent an important way for enhancing of business performance in transition economies.To say distinctly, our findings indicate that in transition economies trade credits positively affect the business performance of firms. Specifically, trust-based relations are associated with increased sales. They provide incentives for more intensive innovation activities and ensure higher labor productivity rates. The firms that trust their partners are characterized by larger proportions of reinvested profits as well.  The main explanation of these findings is that developing of trust among economic agents allows for reduction of transaction costs, stimulates learning and continuous improvement; makes incentives for innovative activities and thus it helps firms in enhancing of their overall business performance. Though, trust-based relations always contain a potential threat of inefficiencies that can arise when low-cost new entrant is excluded, our results suggest that ultimately such relationships are beneficial for firms in transition. The data used in the analysis is well-balanced that makes the results of the study more reliable. However, the sensitivity test indicates that while the estimated effects of trade credit on some indicators of business performance is quite robust, its impacts on the other outcomes are rather sensitive to hidden bias. Another limitation of the study is that it employs only one proxy for trust-based relations and a limited number of performance outcomes indicators.

Thus, for the future research, we propose to investigate the casual links between trust-based relationships and business performance using alternative methods, including instrumental variables technique; employing various proxies for trust and diverse outcome variables. This will allow to supplement the propensity score matching procedure used in this study and to verify the robustness of our findings.

References.

1. Arrow, K.J. 1974. The Limits of Organization. New York: Norton.
2. Becker, S.O., and M. Caliendo. 2007. Sensitivity analysis for average treatment effects, The Stata Journal, volume 7, № 1, pp. 71-83.
3. Berulava G. 2013. Do Trust-Based Relations Improve Firm's Performance? Evidence from Transition Economies, MPRA paper № 48430, July 2013. http://mpra.ub.uni-muenchen.de/49363/
4. Berulava G. and D. Lezhava. 2008. The Impact of Trust on the Mode of Transaction Governance between Manufacturer and Distributor: Evidence from Georgia, Eurasian Journal of Business and Economics, volume 1, № 2 , pp.1-32.
5. Bradach, J.L. and R.G. Eccles. 1989. Price, Authority, and Trust: From Ideal Types to Plural Forms, Annual Review of Sociology, volume 15, pp. 97–118.
6. Carlin W., M. E. Schaffer and P. Seabright. 2004. A Minimum Rivalry: Evidence from Transition Economies on the importance of Competition for Innovation and Growth, William Davidson Institute, working paper № 670.
7. DiPrete, T. and Gangl, M. 2004. Assessing bias in the estimation of causal effects: Rosenbaum bounds on matching estimators and instrumental variables estimation with imperfect instruments, Sociological Methodology, volume 34, №1, pp. 271-310.
8. Dyer, J. H. 1996. Does Governance Matter? KeiretsuAlliances and Asset Specificity as Sources of Japanese Competitive Advantage,  Organization Science, volume 7, № 6, pp: 649-666.
9. Granovetter, M. 1985. Economic Action and Social Structure: The Problem of Embeddedness, The American Journal of Sociology, volume 91, № 3, pp. 481–510.
10. Heide, J. B. and G. John. 1992. Do Norms Matter in Marketing Relationships? Journal of Marketing, volume 56, № 2, pp. 32-44.
11. Hendley, K., P. Murrell, and R. Ryterman. 1998. Law, Relationships, and Private Enforcement: Transactional Strategies of Russian Enterprises, WDI Working Paper No.72, November.
12. Johnson S., J. McMillan and C. Woodruff. 1999. Contract Enforcement in Transition, CESifo Working Paper Series, WP №211.
13. Johnson S., J. McMillan and C. Woodruff. 2002. Property Rights and Finance, American Economic Review, volume 92, № 5, pp. 1335-1356.
14. Lewis, D.J. and A. Weigert. 1985. Trust as a Social Reality, Social Forces, volume63, № 4, pp. 967–985.
15. Macaulay, S. 1963. Non-Contractual Relations in Business: A Preliminary Study, American Sociological Review, volume 28, № 1, pp. 55–67.
16. Mantel, N. and W. Haenszel. 1959. Statistical Aspects of the Analysis of Data from Retrospective Studies of Disease, Journal of the National Cancer Institute, volume 22, № 4, pp. 719-748.
17. McMillan, J. and C. Woodruff. 1999.  Interfirm Relationships and Informal Credit in Vietnam, Quarterly Journal of Economics, volume 114, № 4, pp. 1285–1320.
18. North, D. C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press.
19. Raiser, M., R. Allan, and F. Steves. 2004. Trust in Transition: Cross Country and Firm Evidence, Working Paper No.640 (Ann Arbor: The William Davidson Institute, University of Michigan).
20. Rosenbaum, P. R., and D. B. Rubin. 1983. The Central Role of Propensity Score in Observational Studies for Causal Effects, Biometrika, volume 70, № 1, pp. 41-55.
21. Rosenbaum, P. R. 2002. Observational Studies. New York: Springer.
22. Sako, M. 2002. Does Trust Improve Business Performance? in: Christel Lane and Reinhard Bachmann, eds., Trust Within and Between Organizations: Conceptual Issues and Empirical Applications (Oxford: Oxford University Press).
23. World Bank. 2001. World Development Report 2002: Building Institutions for Markets. The World Bank, Washington DC.
24. World Bank. 2002. Transition – The First Ten Years: Analysis and Lessons for Eastern Europe and the Former Soviet Union. The World Bank, Washington DC.

---