Trade liberalization, profitability, and financial leverage

Abstract

We investigate the effects of trade liberalization on profitability and financial leverage using Canadian data arising from implementation of the Canada–US Free Trade Agreement. We find that falling domestic tariffs are associated with declining profits, especially for import-competing firms, while falling foreign tariffs are associated with increasing profits, especially for export-oriented firms. Also, import tariff reductions tend to increase leverage while export tariff reductions tend to decrease leverage.

Data appendix

The T2-LEAP data set is created by linking the Longitudinal Employment Analysis Project (LEAP) with the Corporate Tax Statistical Universe File (T2SUF). Firms enter the LEAP database when they first hire employees, and exit the database when they cease to have employees. Annual employment for each firm is measured in average labor units (ALU). The reported ALU is interpreted as the number of ‘standardized employees’ working for a firm during a year. A standardized employee corresponds to the industry-specific average (based on payroll data) across full-time and part-time workers.

The T2SUF tracks every incorporated firm in Canada filing a T2 form with the Canada Revenue Agency (CRA). Thus, the T2-LEAP data set contains every firm in Canada that is both incorporated and legally hires employees. We limit our data to firms with more than one employee. This removes the very smallest firms and a lot of ‘noise’ from the data. The eliminated firms are significant in number but negligible in economic importance.

A second filter relates to leverage as measured by debt to asset ratio. Firms report assets, debt, and equity. CRA reporting imposes the constraint that the sum of debt plus equity equals assets. Firms normally determine a book value of assets and a book value of debt according to tax law and generally accepted accounting principles (GAAP). Equity is then determined as the difference between assets and debt. Assets almost always (i.e., in about 97% of cases) exceed debt, leading to a leverage ratio between 0 and 1. However, firms can have debts exceeding the book value of assets, implying a leverage ratio exceeding 1. Most such cases reflect a measurement problem. A firm might borrow money on the basis of a business idea that is, conceptually, an economic asset, but that is not reported as an asset. Debt could then exceed reported assets, and the leverage ratio could exceed 1, even though the ‘true’ debt to asset ratio would be less than 1. In fact, a few firms report positive debt and no assets, leading to infinite leverage ratios. There are also some finite but absurdly large ratios that would be misleading outliers in any regressions. We eliminate all observations for which the debt to asset ratio exceeds 2. This filter eliminates observations whose values consist primarily of measurement error.

T2-LEAP contains firm information for 15 years, from 1984 to 1998. However, the first and last years are subject to partial reporting, leaving the usable portion as 1985 to 1997. We use observations from 1989 forward, and use earlier data when necessary in constructing lags. For each firm, we discard the first and last year of its life in T2-LEAP, as the first and last years will typically be partial years. As we use first differences for some variables, we need two full calendar years of data for a given observation. For example, the firms appearing in our sample for 1989 are those that became incorporated and hired one or more employees on or before 31 December 1987, and did not exit the market before 1990.

The data set includes Canadian subsidiaries of foreign corporations. A large majority of firms either have purely Canadian or widely dispersed ownership. The share of Canadian manufacturing assets controlled by wholly owned or partially owned subsidiaries of American firms is fairly large (approximately 26%), but this ownership is concentrated in large firms (GM Canada, Ford Canada, etc.). We believe that subsidiaries will, in any case, normally be subject to the same incentives as other firms in making capital structure decisions, and that any deviations would have no systematic effect that would bias our analysis.

For firms involved in mergers, acquisitions or spinoffs during the sample period, the T2LEAP record is defined by retrospective reconstruction. If, for example, firm A merged with firm B in year t, then a new firm, C, is created and given a synthetic history aggregated from the histories of firms A and B. The individual histories of A and B disappear from the database, and firm C represents their joint operations.

Using three-digit SIC codes, we are able to match both Canadian and US tariff rates to each firm by year and industry as in Head and Ries (1999). The matching of tariffs to three-digit SIC codes is from Lester and Morehen (1987). The starting point is statutory commodity-level tariffs. These tariffs are aggregated to the three-digit level by taking a weighted average of the underlying commodities in each three-digit category. The weights are production weights where possible, augmented by trade weights. US tariffs are compiled using the 93 industry classification provided in Table A2.1 of the Canada–US Free Trade Agreement: An Economic Assessment (Government of Canada, Department of Finance, 1988).

The annual prime rate and exchange rate are used as control variables. The prime rate was 13.3% in 1989, peaked at 14.1% in 1990, and reached its lowest point in 1997 at 5.0%. The exchange rate was 0.845 US dollars per $C in 1989, rose as high as 0.873 (in 1991), and fell to 0.722 as of 1997.