In this study, I examine whether customer-supplier relationship affect firm's cost structure. Anderson et al. (2003) show that sticky cost occurs because of deliberate managerial decision to adjust the committed resources. I find that cost behavior is less sticky for suppliers with more concentrate-customer base. Additional analysis shows that for suppliers with high level of customer concentration, suppliers' costs are less sticky
when customer's sales decrease. For sample of low level of customer concentration, suppliers’ costs do not response to customer's sales change. The results provide evidence
that supplier-customer relationship has effect on firm's cost structure. My result implies that when supplier have few major customers, there are more information transfers along the supply chain and managers of suppliers are more certain about future demand; thus the stickiness of cost is reduced.
This study contributes to the literature in several ways. Chang et al. (2014) find negative association between cost elasticity and customer concentration. I provide new evidence that customer concentration is negatively associated with cost stickiness. In addition, I also show that information transfers are strengthened when suppliers have high level of customer concentration. Ak and Patatoukas (2016) suggest that collaboration along supply chain helps suppliers increase their inventory efficiency. My result supports
44
their view by showing that this collaboration helps suppliers know more about future demand, adjust their costs timely, and thus reduce the stickiness of costs.
One way forward is to consider the effect of customer sales change in multiple periods. For example, Banker et al. (2014) provide a two-period to test asymmetric cost behavior. Future studies can try to find the association along multiple periods. Another way is to examine whether customer event will affect suppliers cost decision. My study only provides the evidence on customers' sales change. Future studies can incorporate other determinants to have better understanding about firms' cost behavior.
45
References
Ak, B. K., & Patatoukas, P. N. (2016). Customer‐base concentration and inventory efficiencies: Evidence from the manufacturing sector. Production and Operations Management, 25(2), 258-272.
Anderson, M., Banker, R., Huang, R., & Janakiraman, S. (2007). Cost behavior and fundamental analysis of SG&A costs. Journal of Accounting, Auditing &
Finance, 22(1), 1-28.
Anderson, M. C., Banker, R. D., & Janakiraman, S. N. (2003). Are selling, general, and administrative costs “sticky”? Journal of Accounting Research, 41(1), 47-63.
Balakrishnan, R., & Gruca, T. S. (2008). Cost stickiness and core competency: A note.
Contemporary Accounting Research, 25(4), 993-1006.
Balakrishnan, R., Labro, E., & Soderstrom, N. S. (2014). Cost structure and sticky costs. Journal of Management Accounting Research, 26(2), 91-116.
Balakrishnan, R., Linsmeier, T. J., & Venkatachalam, M. (1996). Financial benefits from JIT adoption: Effects of customer concentration and cost structure. Accounting Review, 183-205.
Balakrishnan, R., Petersen, M. J., & Soderstrom, N. S. (2004). Does capacity utilization affect the “stickiness” of cost? Journal of Accounting, Auditing & Finance, 19(3), 283-300.
46
Banker, R. D., & Byzalov, D. (2014). Asymmetric cost behavior. Journal of Management Accounting Research, 26(2), 43-79.
Banker, R. D., Byzalov, D., & Chen, L. T. (2013). Employment protection legislation, adjustment costs and cross-country differences in cost behavior. Journal of Accounting and Economics, 55(1), 111-127.
Banker, R. D., Byzalov, D., Ciftci, M., & Mashruwala, R. (2014). The moderating effect of prior sales changes on asymmetric cost behavior. Journal of Management Accounting Research, 26(2), 221-242.
Banker, R. D., & Chen, L. (2006). Predicting earnings using a model based on cost variability and cost stickiness. The Accounting Review, 81(2), 285-307.
Calleja, K., Steliaros, M., & Thomas, D. C. (2006). A note on cost stickiness: Some international comparisons. Management Accounting Research, 17(2), 127-140.
Cen, L., Dasgupta, S., Elkamhi, R., & Pungaliya, R. S. (2016). Reputation and loan contract terms: The role of principal customers. Review of Finance, 20(2), 501-533.
Chang, H., Hall, C., & Paz, M. T. (2015). Customer concentration and cost structure.
Available at SSRN 2482777
Chen, C. X., Gores, T., & Nasev, J. (2013). Managerial overconfidence and cost stickiness. Available at SSRN 2208622.
47
Chen, C. X., Lu, H., & Sougiannis, T. (2012). The agency problem, corporate governance, and the asymmetrical behavior of selling, general, and
administrative costs. Contemporary Accounting Research, 29(1), 252-282.
Chen, J., Chang, H., Chen, H.-C., & Kim, S. (2014). The effect of supply chain knowledge spillovers on audit pricing. Journal of Management Accounting Research, 26(1), 83-100.
Cohen, L., & Frazzini, A. (2008). Economic links and predictable returns. The Journal of Finance, 63(4), 1977-2011.
Cool, K., & Henderson, J. (1998). Power and firm profitability in supply chains: French manufacturing industry in 1993. Strategic Management Journal, 19(10), 909-926.
Cowley, P. (1988). Market structure and business performance: An evaluation of buyer/seller power in the PIMS database. Strategic Management Journal, 9(3), 271-278.
Dierynck, B., Landsman, W. R., & Renders, A. (2012). Do managerial incentives drive cost behavior? Evidence about the role of the zero earnings benchmark for labor cost behavior in private Belgian firms. The Accounting Review, 87(4), 1219-1246.
Galbraith, J. K. (1952). American capitalism: The concept of countervailing power:
48
Boston, MA: Houghton Mifflin.
Gosman, M. L., & Kohlbeck, M. J. (2009). Effects of the existence and identity of major customers on supplier profitability: is Wal-Mart different? Journal of Management Accounting Research, 21(1), 179-201.
Guan, Y., Wong, M. F., & Zhang, Y. (2015). Analyst following along the supply chain.
Review of Accounting Studies, 20(1), 210-241.
Gulati, R., & Sytch, M. (2007). Dependence asymmetry and joint dependence in interorganizational relationships: Effects of embeddedness on a manufacturer's performance in procurement relationships. Administrative science quarterly, 52(1), 32-69.
Hertzel, M. G., Li, Z., Officer, M. S., & Rodgers, K. J. (2008). Inter-firm linkages and the wealth effects of financial distress along the supply chain. Journal of Financial Economics, 87(2), 374-387.
Huang, H. H., Lobo, G. J., Wang, C., & Xie, H. (2015). Customer concentration and corporate tax avoidance. Available at SSRN 2469859.
Irvine, P. J., Park, S. S., & Yildizhan, Ç. (2015). Customer-base concentration,
profitability, and the relationship life cycle. The Accounting Review, 91(3), 883-906.
Kale, J. R., & Shahrur, H. (2007). Corporate capital structure and the characteristics of
49
suppliers and customers. Journal of Financial Economics, 83(2), 321-365.
Kalwani, M. U., & Narayandas, N. (1995). Long-term manufacturer-supplier
relationships: do they pay off for supplier firms? The Journal of marketing, 1-16.
Kama, I., & Weiss, D. (2013). Do earnings targets and managerial incentives affect sticky costs? Journal of Accounting Research, 51(1), 201-224.
Kim, J.-B., Song, B. Y., & Zhang, Y. (2015). Earnings performance of major customers and bank loan contracting with suppliers. Journal of Banking & Finance, 59, 384-398.
Kim, M., & Prather-Kinsey, J. (2010). An additional source of financial analysts' earnings forecast errors: Imperfect adjustments for cost behavior. Journal of Accounting, Auditing & Finance, 25(1), 27-51.
Krishnan, G. V., Lee, H. S., Patatoukas, P. N., & Wang, A. Y. (2015). Major customer dependency: Implications for audit pricing and quality. Available at SSRN 2440009.
Kumar, N. (1996). The power of trust in manufacturer-retailer relationships. Harvard business review, 74(6), 92-106.
Malik, M. (2012). A review and synthesis of'cost stickiness' literature. Available at SSRN 2276760.
Noreen, E., & Soderstrom, N. (1994). Are overhead costs strictly proportional to
50
activity?: Evidence from hospital departments. Journal of Accounting and Economics, 17(1), 255-278.
Noreen, E., & Soderstrom, N. (1997). The accuracy of proportional cost models:
evidence from hospital service departments. Review of Accounting Studies, 2(1), 89-114.
Olsen, C., & Dietrich, J. R. (1985). Vertical information transfers: The association between retailers' sales announcements and suppliers' security returns. Journal of Accounting Research, 144-166.
Pandit, S., Wasley, C. E., & Zach, T. (2011). Information externalities along the supply chain: The economic determinants of suppliers’ stock price reaction to their customers’ earnings announcements. Contemporary Accounting Research, 28(4), 1304-1343.
Patatoukas, P. N. (2011). Customer-base concentration: Implications for firm performance and capital markets. The Accounting Review, 87(2), 363-392.
Porter, M. E. (1974). Consumer behavior, retailer power and market performance in consumer goods industries. The Review of Economics and Statistics, 419-436.
Porter, M. E. (2008). The five competitive forces that shape strategy. Harvard business review, 86(1), 78-93.
Qin, B., Mohan, A. W., & Kuang, Y. F. (2015). CEO overconfidence and cost stickiness.
51
Management Control & Accounting (2), 26-32.
Raman, K., & Shahrur, H. (2008). Relationship-specific investments and earnings management: Evidence on corporate suppliers and customers. The Accounting Review, 83(4), 1041-1081.
Ren, Z. J., Cohen, M. A., Ho, T. H., & Terwiesch, C. (2010). Information sharing in a long-term supply chain relationship: The role of customer review strategy.
Operations research, 58(1), 81-93.
Scherer, F. M. (1970). Industrial Market Structure and Economic Performance. Chicago, IL: Rand McNally.
Shahrur, H., Becker, Y. L., & Rosenfeld, D. (2010). Return predictability along the supply chain: the international evidence. Financial Analysts Journal, 66(3), 60-77.
Subramaniam, C., & Weidenmier, M. L. (2003). Additional evidence on the sticky behavior of costs. Social Science Research Network.
Weiss, D. (2010). Cost behavior and analysts' earnings forecasts. The Accounting Review, 85(4), 1441-1471.
Yasukata, K. (2011). Are 'sticky costs' the result of deliberate decision of managers?
Available at SSRN 1444746.
52 Appendix 1 : Variable definitions
Variable definitions
ln
Log change operatorSales Sales Revenue in million (Compustat item SALE)
SGA Selling, general and administrative expense (Compustat item XSGA)
OC Total operating costs [revenue minus operating income] (Compustat item XOPR) COGS Cost of goods sold (Compustat item COGS)
CC Customer concentration score for firm i in year t (
CC
it ) equalsDeci.t 1 if sales revenue from firm i decreased in year t relative to year t-1, 0 otherwise.
CDeci.t 1 if customer’s sales revenue from firm i decreased in year t relative to year t-1, 0 otherwise.
GDPGrowth GDP growth in year t.
Source: http://www.bea.gov/national/xls/gdpchg.xls Size Natural log of sales
ASINT Asset intensity, defined as the log-ratin of number of employees to sales.
ln(AT/SALE)
EMPINT Employment intensity, defined as the log-ratio of number of employees to sales.
ln(EMP/SALE)
53
Appendix 2: Models of cost asymmetry
Noreen and Soderstrom (1994) noted that the proportional cost model assumes that cost is strictly proportional to single measure of activity. The cost model assumes
C = p ∗ q (1)
where p is a positive constant, and q is the activity measure. However, the model results in heteroscedastic residuals. In order to address this problem, the following logarithmic form is used
ln(C) = ln(p) + βln(q) (2)
The slope coefficient β is the ratio of marginal cost to average cost.4 In other word, β evaluates how much a given percentage of change in activity turn into a percentage change in cost. Thus, in this logarithmic form, the test of whether cost is proportional to activity is simplified to a test whether β is 1. If β = 1, the cost model is consistent with the proportional cost model. If β < 1, the cost model is consistent with increasing return of scale.
In 1997, Noreen and Soderstrom test whether proportional cost model is accurate.
They start from activity-based costing model. Costs, resources and activities are linked in a very specific way under activity-based costing. Single period ABC model is
TCt = ptαqt (3)
4 If C = p ∗ qβ, Average cost (AC) is C/q = p∗qβ
q = p ∗ qβ−1, and marginal cost (MC) is β ∗ p ∗ qβ−1. = βAC , β = MC/AC
54
where
TC = total cost in a cost pool in period t
p = the price per unit of the overhead resource in period t
α = the amount of the overhead resource consumed per unit of activity q = activity in period t
The model assumes that α is constant, while the price pt can vary from one period to the next.
Using the above model, costs can be estimated from prior period data and from anticipation of pt and qt as follows: changes in activity level and changes in input prices. Since they are interest in the change in activity level, the actual change in cost after adjust for inflation is:
1
The estimated change in cost after adjust for the changes in prices is:
1
Then, they standardize the estimation error in the following way:
1
55
Zt is the percentage error in estimating the change in total price. If the proportional model predicts the change correctly, Zt is zero. Negative value of Zt indicates that the proportional model underestimated total costs, while positive value of Zt indicates that the model over estimated total costs.
Next, Noreen and Soderstrom show multi-period ABC model:
(q )
C(qƞ) = cumulative undiscounted cost consequences over all future period of qƞ
α = amount of resources consumed per unit of activity
= percentage of resource consumption whose cost is realized in the period subsequent to the change in activityp
= the price per unit of resource in period τHere, the assumption is quite similar to one-period model. The assumption here is that cost may not occur in the current period, but it will finally realize.
The realized total cost for a particular cost pool in period t, TCt, is the result of activity in current period and previous T period as follows:
56
Next, take the ratio of current period’s total cost to previous period’s total cost:
1
Using the above model, predicted total cost can be calculated by the following model:
1
0 1 1The results of one period model and multiple-period model show that overhead cost pool is not strictly proportional to activity, thus Noreen and Soderstrom, taking logs
on both side of multiple-period model, provide the following model:
Assume the ratio of the successive activity levels are close to one, equation 11 can be rewritten in approximation form:
57 response of SG&A costs to contemporaneous change in sales revenue and the different between periods when sales increase and sales decrease is presented:
, , , compared with prior period. They take the model in ratio and logarithmic form to improve the comparability of the variable across firms and reduce the effect of potential heteroskedasticity, which has been mentioned in Noreen and Soderstrom's model.
The model provides the basis for test of cost stickiness. If traditional cost model is correct, upward and downward changes in costs will be equal, thus β2 = 0. Besides, if fix costs shows, then β1 is smaller than one, signifying economic of scale .The coefficient β1
measures percentage increase in SGA cost when sales increase in 1 %. Since the
_
Decrease Dummy is one when sales decrease, β1 + β2 measures the percentage increase in SGA cost when sales decrease in 1 %. If costs are sticky, the change in cost when sales increase should be greater than the change when sales decrease. Thus, the hypothesis for stickiness, conditional on β1 >0 , is β2 <0. The hypothesis of asymmetric cost behavior is similar to the model proposed by Noreen and Soderstrom.
Balakrishnan et al. (2014) suggest that the asymmetric behavior of costs is conditional on the proportion of fixed costs. They propose a model to alleviate the effect of fixed costs structure on the asymmetric response. They begin with Anderson et al.(2003)’s model:
58
Dec = indicator variable equal to 1 if sales decline from period t-1 to t, and 0 otherwise.
To make it simple, they begin by considering a single firm with two periods of activity. Consider a firm with the following cost structure:
t * t
Within this setting, the percentage change for cost and revenues is:
1
59
Ignore the effect of cost asymmetric, and their linear cost equation is:
CstSale(St-1 , Linear) = 1
The ratio of variable to total costs increases in revenue because fixed costs are
spread over large volumes. In other words,
0
t
CstSale(St-1) is monotonically increasing at decreasing rate.
Next, suppose there are F identical firms with same cost structure and sales volume. Then assume that all of the firms’ sales increase by 10% from period t-1 to period t, except some proportion ρ has a decline in sales rather than increase. And the next year (i.e., from t-1 to t+1), all of the firms have a 10% increase in sales.
Consider the following equation on the two changes:
1
*
. They conclude that the greater proportion of firmswith a decline in activity, the greater the downward bias affected in the estimate of
1.For this reason, they argue that the presence of fixed cost negatively biases the
estimated coefficient of
2 away from zero, thus reject the null hypothesis of no asymmetry. They propose an alleviated model to deal with the problem. They scale the model by lagged sales activity. The model is60
And percentage change for cost and revenues become:
1
The cost response to the scaled changes in sales is then:
CstSale(St-1 , Linear) =
The proportion of fixed costs to total costs and the level sales of current period will no longer bias the result. The results can be reasonably interpreted as managerial
decision.
Banker et al. (2014) refine the theory and empirical model proposed by Anderson et al.(2003). Here again the original model is:
, , ,
Banker et al. (2014) argue that cost asymmetry is more complex than Anderson et el.(2003) said. The original model shows the average degree of cost asymmetry, but fail to distinguish between the two underlying processes of conditional stickiness and anti-stickiness. They proposed the following two-period and three-period model:
Two-period model:
61 if sales increase (decrease) from period t-2 to period t-1, and 0 otherwise. The
coefficient of
1PIncrand
2PIncr(
1PDecrand
2PDecr) correspond to
1 and
2 in the original model as the subsample of observation that follow a prior sales increase (decrease).The modified model with two and three period reflects to the effects of prior sales change, and the results show that cost stickiness conditional on a prior sales increase, and cost anti-stickiness conditional on a prior sales decrease.
Weiss (2010) proposed a new measure of cost stickiness at the firm level. Most studies on cost stickiness use a cross-sectional model or a time-series model to estimate cost stickiness. He creates a direct measure of cost stickiness at the firm level. He estimate the difference between the rate of cost decrease for recent quarter with sales decrease and the corresponding rate of cost increase for recent quarters with increasing
sales in the following equation:
62
where
= the most recent of the last four quarters with a sales decrease
= the most recent of the last four quarters with a sales increase.SALE
it
= Saleit - Saleit-1COST
it
= (Saleit - EARNINGSit) - (Saleit-1 - EARNINGSit-1)Earnings = Income before extraordinary items
STICKY is defined as the difference between the two most recent quarters from t-3 to to quarter t that sales decrease in one quarter and increase in the other. If cost are sticky, cost increase more when activity rise than they decrease when activity fall by the same volume. Therefore, STICKY will have a negative value. The lower value of STICKY means the costs are more sticky. In other words. negative (positive) value of STICKY shows that managers are less(more) likely to adjust the costs when sales decrease than they are when sales increase by the same amount.
63
Table 1: Descriptive statistics of test variables in H1
Percentiles
Variable N Mean Std Dev 25th 50th 75th
SALE 47,452 834.0031 2395.7024 20.2625 90.6185 450.7840
OC 47,452 700.7037 1968.5846 20.3945 82.7415 391.1820
SGA 47,452 150.6157 439.0183 5.5400 20.3060 80.6670
COGS 47,452 527.4510 1502.6128 11.9610 55.1570 280.2820
EMP 47,452 4.3292 16.0825 0.1420 0.5540 2.5000
CC 47,452 0.1043 0.1440 0.0169 0.0507 0.1308
SIZE 47,452 4.5701 2.2294 3.0088 4.5067 6.1110 ASINT 47,452 1.4181 35.4726 0.6399 0.8743 1.2924 EMPINT 47,452 0.0089 0.0113 0.0037 0.0064 0.0110
64
Table 2: Correlation table
Var SALE SGA OC COGS CC GDPGrowth Size ASINT EMPINT
SALE 1 0.84995 0.99426 0.96228 -0.08015 -0.08966 0.60645 -0.00265 -0.13297
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.5637 <.0001
SGA 0.89453 1 0.83233 0.72874 -0.09139 -0.09465 0.55766 0.00066 -0.1301
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.8852 <.0001
OC 0.99414 0.91066 1 0.97728 -0.08242 -0.09013 0.61164 -0.00255 -0.13239
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.579 <.0001
COGS 0.9865 0.84347 0.98365 1 -0.08071 -0.08585 0.60285 -0.00348 -0.12772
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.448 <.0001
CC -0.18774 -0.17832 -0.17883 -0.18606 1 -0.04552 -0.1989 0.02456 0.03405
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
GDPGrowth -0.14184 -0.17787 -0.15169 -0.12741 -0.07805 1 -0.1096 -0.00397 0.09297
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.3873 <.0001
Size 1 0.89454 0.99414 0.9865 -0.18777 -0.14184 1 -0.02317 -0.31216
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
ASINT 0.00761 0.15183 0.01926 -0.05318 0.07811 -0.10696 0.00762 1 0.5135
0.0975 <.0001 <.0001 <.0001 <.0001 <.0001 0.097 <.0001
EMPINT -0.41889 -0.42959 -0.41346 -0.37642 -0.0648 0.23638 -0.41891 -0.05847 1
<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
65
Table 3: Main result for H1-SGA
ln(SGA)
ln(SGA) ln(SGA)
Intercept 0.04295 -0.01586 -0.01958
(21.49)*** (-3.72)*** (-2.57)**
66
Table 4: Main result for H1-COGS
ln(COGS)
ln(COGS) ln(COGS)
Intercept -0.00981 -0.01958 -0.02494
(-5.45)*** (-5.07)*** (-3.57)*
67
Table 5: Main result for H1-OC
ln(OC)
ln(OC) ln(OC)
Intercept 0.02586 -0.0604 -0.02936
(19.14)*** (-9.16) (-5.88)***
68
Table 6: Descriptive statistics of test variables in additional test
Percentiles
Variable N Mean Std Dev 25th 50th 75th
SALE 15028 1312.8250 3739.7376 30.8225 142.7255 721.7535
OC 15028 1111.8204 3109.0992 35.6905 135.1195 630.2910
SGA 15028 248.6158 713.2046 8.0385 33.0620 134.4415
COGS 15028 822.0709 2373.1189 21.7285 92.4950 462.2560
EMP 14637 5.8050 18.3629 0.1830 0.7280 3.3040
SIZE 15028 5.0042 2.2712 3.4282 4.9609 6.5817 ASINT 15026 3.9830 77.2109 0.6744 0.9624 1.5546 EMPINT 14637 0.0099 0.0590 0.0030 0.0051 0.0095
69
Table 7: Subsample Analysis-Impact of customer’s sales decrease to cost stickiness-SGA
ln(SGA) -L ln(SGA)-H ln(SGA)-L ln(SGA)-H ln(SGA)-L ln(SGA)-H
Intercept 0.06073 0.04496 0.02565 0.05125 -0.00274 0.05785
(10.34)*** (11.62)*** (2.01) (5.09)*** (-0.13) (3.39)***
GDPGrowth 0.51801 0.22801 0.57367 0.25350
(2.73)*** (1.73)* (3.04)*** (1.94)*
Size -0.00081 -0.00621 0.00176 -0.00460
(-0.48) (-5.30)*** (1.00) (-3.75)
ASINT 0.00034 0.00453 -0.00006 0.00548
(1.26) (3.64)*** (0.24) (4.41)***
EMPINT 2.04578 0.97716 2.29228 1.25597
(6.41)*** (3.17)*** (7.15)*** (3.97)***