We then estimate total factor productivity at the firm level by taking into account the measured probabilities of binding financial constraints. Hennessy and Whited (2007) noted that the Kaplan-Zingales index of financial constraints measures the firm's “need” for external financing, rather than the degree to which it is constrained by credit market frictions. In this paper, the data allow us to separate financial constraints from the need for external financing.
Neither the underlying productivity process nor the degree of economic constraints are estimated in these models. Instead, financial constraints are approximated by a set of observable values that are assumed to indicate the firm's (in)ability to borrow. Because of this, our measure of financial constraints may underestimate the degree of a firm's financial constraint.
Variables that affect how binding financial constraints are, fj t, should also affect the demand for external financing.
4 Financial Constraint and Productivity
- Production function estimation with financial constraint
- Estimation procedure
- Diagnostics
- Production function estimation with financial constraint
- Bias of estimated productivity
- Financial constraint and endogenous productivity
If financial constraints bind, investment and employment are likely to be lower than otherwise, conditional on the realization of a productivity shock. Whether we can consider financial constraints in productivity estimation in the spirit of Olley-Pakes depends on how the financial constraints are specified. However, we can use the measured likelihood of being financially constrained in the estimation, allowing us to isolate financial constraints from productivity in shaping investments. In addition, selection bias due to endogenous firm exit is associated with financing.
The second stage of estimation in the Olley-Pakes method should then use variables for financial constraints to estimate the probability of firm exit.5. We first examine the correlation between financial constraints and productivity based on the estimated production function without considering financing. Second, the negative correlation between estimated productivity and measured financial constraints also suggests that productivity may influence the estimate of the probability of being financially constrained.
In the case of exogenous productivity, the estimates of labor and capital coefficients are larger than the estimates when external financing and financial constraints are omitted. This is expected, since the correlation between financial constraints and labor and investment choices is negative. This is probably due to the way we measure the likelihood of binding financial constraints.
The estimation also used proxy variables for total factor productivity, such as firm size and debt, to assess the likelihood of financial constraints. Estimated total factor productivity (de-meaned,ωj t−ω) and measured financial constraints are still negatively correlated, with a correlation coefficient of -0.45, stronger than the -0.30 obtained in the case of the no-finance estimate. To diagnose whether the measured financial constraints affect productivity, we estimate the production function with external financing (debt), but assume that there is no financial constraint, so in the first stage of the ACF estimation we use ωj t=M−1(kj t,bj t ,lj t,mj t) to approximate productivity.
However, when the production function is estimated with funding and financial constraints (that is, ωj t =M−1(kj t,bj t,lj t,mj t,fj tc) is used in the first stage of ACF), estimates of γ4 are all positive and mostly not statistically significant (as shown in Table 14). This suggests that, once the estimation error due to the omission of financial constraints is corrected, the measured probability of constraints no longer has a separate effect on productivity processes.
5 Productivity, Financial constraint and Firm growth
Table 15 shows that companies with a higher lagged productivity and a higher probability of disability are less productive, resulting in lower productivity growth. In summary, we do not find a significant estimate of the effect of financial constraints on productivity, after the productivity estimate corrects for the bias due to the omission of financial constraints. It is more likely that companies with limited cash flow are less likely to invest a positive amount.
In addition, the probability of a positive investment increases if the current debt ratio becomes larger. Investments are, as usual, sensitive to cash flows; companies with a higher cash flow are more likely to invest. The total debt to asset ratio is insignificant and its coefficient has an opposite sign if this ratio is assumed to reflect the degree of financial constraint.
In column (2), where only the measured unconditional probability of being financially constrained is included, we observe that the more likely the firm is constrained, the less likely its investment is positive. Table 17 reports results of linear regressions of employment growth on financial variables and measured financial constraints. When TFP estimates are taken into account, it is clear that the degree to which firms are financially constrained continues to have a negative impact on employment growth.
Coefficient estimates for total factor productivity are negative, but statistically insignificant for lagged total factor productivity, similar to results based on sector-level data by Basu, Kimball and Fernald (2006).9 Finally, the expanded investment margin is adversely affected. by the measured degree of financial constraints, and positively affected by total factor productivity, both estimates are statistically significant. 9Not reported here, coefficient estimates for the change in total factor productivity are negative and statistically significant. A potential explanation for the negative effect of total factor productivity on employment growth is nominal rigidity in output prices.
6 Conclusions
Ruggieri (2015): "Financial Constraints and Productivity: Evidence from Euro Area Companies," Working Paper Series 1823, European Central Bank. Zingales (1997): "Giver investerings-pengestrømsfølsomhed nyttige mål for finansieringsbegrænsninger?", The Quarterly Journal of Economics. Warusawitharana (2014): "Finance and Productivity Growth: Firm-level Evidence," Finance and Economics Discussion Series 2014-17, Board of Governors of the Federal Reserve System (U.S.).
Baldwin (2013): "Canadian Labor Productivity Differences Across Firm Size Classes, 2002 to 2008,"The Canadian Productivity Review2013032e, Statistics Canada,.
Appendix A Data Description
SMEs: Some basic facts
Summary statistics on financing, 2011
For those that did not request external financing in 2011, 88 percent of companies reported that there was no need for external financing, about 6 percent reported that the request for financing would be rejected or that it was too difficult (or time consuming) to apply for financing. or the financing costs were too high. In 2011, 25 percent of all businesses requested loans (they could also have requested other financing such as trade credit). In total, about 8 percent of all businesses requested trade credit, and 3.7 percent of all businesses requested some form of public financing.10.
In 2011, the majority of external financing was intended to finance the purchase of land and buildings, vehicles, information technology, equipment and working capital. Among those businesses that applied for loans in 2011, more than 50 percent applied for business lines of credit (new or increased limits), 40 percent applied for a business credit card, 35 percent applied for a long-term loan, and only 16 percent applied for non-residential mortgage (new or refinanced). Note that this sum does not add up to 100, as companies can apply for more than one type of loan.
The approval rate for loan applications is high – in 2011, 85 percent of applications were approved for the full amount, while only 5 percent of applications were approved for a partial amount. In these rejected loan applications, the main reasons for rejection were poor or no credit history and the project was deemed too risky. Rejected loans were not concentrated into one specific loan type.
Please note that some loans involved refinancing or additions to existing loans. A company can therefore report that no collateral was required for the new part of an existing loan.
Appendix B Score Assigning
He does not apply for external funding because: (i) he thought the application would be rejected, (ii) applying for funding is too difficult or time-consuming, or (iii) the cost of funding is too high. Other financing requests (including leases, equity, commercial credit and government financing) were rejected or only partially approved. We then subtract a score from the original score based on how well the company's responses meet the above criteria.
The highest scores are given to those who did not apply for external financing and reported that it was too expensive. Companies that have applied for external financing but do not meet any of the above criteria are given the lowest score. In between are companies that meet some of the above criteria – for example, loans have been partially approved or the company has turned to the state for loans or loan guarantees.
The indication of limitation defined here probably represents a lower bound of the degree of limitation. Firms that have made funding requests and obtained full authorization may still be financially restricted. These firms may have knowledge of the underlying financial frictions and therefore could have taken such knowledge into account and requested an amount that the bank would approve with full authority.
In 2011, almost 43 percent of companies needed external financing, including 7.8 percent of companies that needed financing but did not apply for it. Companies that report no need for external financing have no value for the degree of financial constraint.
Appendix C Data Variable Definition
Total assets: equals current assets plus capital assets (i.e. machinery, equipment, furniture and buildings) plus long-term financial assets (e.g. stocks, loans). Added value: equals total sales minus the cost of intermediate inputs, the latter equals the cost of sales minus wages and crown costs. Hours Worked: Calculated as annual total payroll divided by average hourly wages by region and at the 3-digit NAICS level.
Hourly wages are drawn from the Survey of Employment, Payrolls and Hours (SEPH) by Statistics Canada. When necessary, real variables are obtained by using the current price values divided by corresponding implied prices at the 2-digit NAICS level. These implicit prices are drawn from the multifactor productivity data sets of Statistics Canada.
Appendix D Tables
Note: This is estimated with OLS after obtaining TFP estimates using our ACF procedure where the productivity process is the same as shown in this table. Note: This is estimated using OLS, after obtaining TFP using our ACF procedure, where the productivity process is the same as shown in this table. TFP is estimated in the economic constraint model, assuming that the TFP process is exogenous.
WORKING PAPERS IN ECONOMICS AND FINANCE
