Nominal Rigidities in a Mail Order Company: Estimation of the Probability of Price Adjustment

Bertarelli, Silvia Working Paper Nominal Rigidities in a Mail Order Company: Estimation of the Probability of Price Adjustment Quaderni - Working Paper DSE, No. 349 Provided in Cooperation with: University of Bologna, Department of Economics Suggested Citation: Bertarelli, Silvia (1999) : Nominal Rigidities in a Mail Order Company: Estimation of the Probability of Price Adjustment, Quaderni - Working Paper DSE, No. 349, Alma Mater Studiorum - Università di Bologna, Dipartimento di Scienze Economiche (DSE), Bologna, https://doi.org/10.6092/unibo/amsacta/4960 This Version is available at: https://hdl.handle.net/10419/159190 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-nc/3.0/ Nominal Rigidities in a Mail Order Company: Estimation of the Probability of Price Adjustment by Silvia Bertarelli Dipartimento di Scienze Economiche Università di Bologna E-mail [email protected] Last draft: April 14, 1999 Abstract: Nominal price rigidities are analyzed from two different perspectives. From a qualitative point of view - as Kashyap (1995) - we look for coherent signs which confirm the presence of fixed costs of adjustment, and therefore the validity of state-dependent pricing rules. Our data concern a mail order Italian company and confirm that individual prices do not mimic the aggregate price index, as it was already claimed by Kashyap (1995) and Tsiddon (1993) for US and Israeli data. Second, we estimate by ordered probit models the probability to have nominal price adjustments (increase, rigidity or decrease) as a function of last period real price and rivals’ price level. Price point strategies are also investigated. We observe a common performance for all brands’ prices which supports a (S, s) rule hypothesis. JEL Classification: C25, D40, D43. 2 1. Introduction Our main objective is to verify the consistency of nominal price rigidities with a hypothesis of fixed costs of adjustment by using individual mail order prices. We will present results of an empirical analysis with reference to catalogue prices set by an Italian company of mail order retailing. Nominal rigidity is an important issue from a theoretical and an empirical perspective. The decision to study optimal pricing rules connected with fixed costs of adjustment has been originated by the observation of empirical works describing aggregate and individual prices behavior. In the former case prices are characterized by smoothness and adjustments are continuous and partial. In individual data set, prices adjust very rarely and the variation is not negligible. From a microeconomic point of view, there are several ways to justify stickiness of prices. In the presence of costs to collect information the optimal pricing rule is time-dependent. In the presence of fixed costs of adjustment, i.e. menu costs, the optimal pricing rule is state-dependent. When the observation of the state of the world is costly the optimal firm's decision is to do it at constant intervals of time and therefore the nominal price does not change during these periods. Taylor (1979, 1980) and Fischer (1977) originally proposed models with time-dependent pricing strategies. In order to find evidence confirming the theory, we have analyzed the length of all periods of rigidity1. Kashyap (1987) study on mail order prices of 3 different US mail-order companies shows the inconsistency of time-dependent rules even though claiming that these rules are a good justification in case of small price changes. However, a fixed time dependent pricing rule is not sufficiently general. Kashyap (1995) observes that the length of the period of rigidity is related to the information released by the economic system over time. Thus a variable time-dependent rule is preferred. If two goods are similar as to demand and costs characteristics, we should find a strong correlation between their timings in the adjustment2. In our data set, however, we do not find a high degree of correlation and association. This result is not surprising since they refer to very different 1To check the existence of fixed intervals of rigidity we have determined the average and median rigidity duration and its standard deviation. We can say that there are no constant intervals of rigidity. Even taking into account inflation dynamics, we do not find higher of adjustment frequencies in the high inflation period (1973-1984) than in the low inflation one (1985-1995). We instead see that prices were adjusted less frequently in the two-digit inflation regime than in the other one. Therefore we cannot accept a model where prices remains constant for given intervals of time. 2 To verify if price adjustments of two items are synchronized we calculate a measure of association, using a string of (-1, 0, +1) for each product. This variable indicates with "-1" if a price decreased with respect to the previous semester, "0" if it was unchanged, and "+1" if it increased. We can also determine a correlation index between two goods’ price change rates. In this case, we take into account variations’ magnitude and sign. 3 items for which it is difficult to admit common features about demand and cost conditions3. It is therefore impossible to give a final answer to our question about the importance of time dependent rules. Since our data are not suitable to check the goodness of time dependent rules, we will restrict our attention to state dependent rules and we will try to estimate the probability of adjustment in the hypothesis of this pricing policy. The optimality of state-dependent rules is demonstrated in dynamic models where a firm pays a menu-cost to adjust nominal prices. The optimal decision establishes the variation of the nominal price when the real price is too far from the optimal real price, whose level refers to an assumption of adjustment at no cost. The state variable - the real price - fluctuates between the optimal price and the upper or the lower limit of a band of oscillation. Menu costs can be interpreted as explicit costs to adjust prices or can represent a good approximation of firms' behavior in models of near rationality, or models with uncertainty and imperfect markets, or in search models4. It will be interesting to verify the consistency of our data set with a state dependent rule in the hypothesis of fixed and variable bands. In this framework we will study some phenomena such as downward rigidity and small price adjustments. Since we are considering all factors influencing the decision to adjust nominal prices, additional stickiness can be introduced if firms have a preference for price points. This idea was suggested by Kashyap (1995) and confirmed by Blinder (1991) in an interview study about price stickiness. In marketing theory a price point is a limit the firm gets over with difficulty and only to reach a new price point. An example of price point in Italian lire is L 19900. The idea is to describe imperfectly rational consumers who have different perception of a price like L 19900 with respect to a little higher price like L 20100. Friedman (1967) gives empirical evidence about a large use of price points in US data. A dummy variable to indicate if a price is a price point or not is important in the 3 For a few products we obtained a quite large association index. The same result is determined when we consider a correlation index. For example, for the couples bedspring-cover and mattress-cover the association index is 0.79 and the correlation is 0.78. Almost the same result is obtained for the couple plastic boxes and bags-holder. These items are very similar and we can think a common pricing policy is applied since the former group is presented in pair, even though it is possible to buy them separately; the latter couple is produced by the same firm and constructed with the same materials. For other brands, we found a certain degree of association but a negative (and close to zero) correlation. For example, for the couple plastic boxes and electric cables the index of association is 0.24 and the correlation index is -0.04. Other couples with this feature are: blood pressure gauge and electric cables, blanket and plastic boxes. In these cases we run the risk that the correlation index give a greater importance to change rates’ magnitude than to their sign. 4 References are to Barro (1972) for a model of optimal pricing with fixed adjustment costs, and to near rationality model by Akerlof and Yellen (1985). Greenwald and Stiglitz (1993) review theoretical works about uncertain and imperfect capital and labor markets. Sequential search models applied to goods’ market theory represent a possible explanation of countercyclical mark-ups (Stiglitz, 1979). 4 determination of the probability of price adjustment; we will see it in our econometric work in section 6. The paper is organized as follows. Some considerations about mail order prices are presented in section 2. Then we consider the analysis of nominal prices (section 3) and real prices (section 4) from a qualitative point of view with the main objective to verify the consistency of state-dependent rules with our data set. Last part of the work (section 5 and 6) is devoted to the estimation of the probability of adjustment in the hypothesis of an ordered probit model associated with a statedependent pricing policy. 2. Mail order prices In a mail order firm a fixed cost of adjustment is represented by the cost of a catalogue which is uncorrelated with the magnitude of the price adjustment. The first paper where a (S, s) rule has been presented as the optimal rule in the presence of fixed adjustment costs, Sheshinski and Weiss (1977), suggests some examples one of which is the mail order case. Several empirical works point out diverging behavior of individual prices and aggregate price indexes. Microeconomic data are studied in some papers like Mussa (1981), Carlton (1986), and Cecchetti (1986) 5. As Kashyap (1995), we are concerned with mail order catalogue prices. We are not interested in finding why a price is fixed for six months6; on the contrary, our final purpose is to explain why we often have the same price for several semesters even though it would be possible to change it on a new catalogue. Moreover, the price is the only term of exchange between the firm and its customers and there are no additional conditions in the contract to define. Long-term contracts are absent even though the firm is interested to transform occasional buyers into regular customers. In this section we use prices of 11 goods collected in Postalmarket catalogues from 1970 to 1995. Appendix gives some information about the company and selected brands. All 11 items have some required characteristics. Quality transformations were limited. We avoided goods with a very short business cycle, fashion brands and new items. All goods are also available in other systems of distribution. On May and November, the firm chooses all prices for the new catalogue. In table 1 we 5Mussa (1981) analyzes prices in the hyperinflation period that stroke Germany during the 20s’. Carlton (1986) considers intermediate goods’ prices and Cecchetti (1986) studies the prices of the main magazines issued in US. 6Some economists give us some useful intuitions about the rigid versus the indexed price issue. Gordon (1990) claims that firms are interested in their demand and cost functions so aggregate demand indexation cannot be acceptable. Weiss (1993) observes that indexation is possible if a lot of firms do the same. This rule has a "public-good"-like nature so multiple equilibria are possible. 5 can observe some statistics as to sample length, per semester average change rates and standard deviations. Moreover, the longest rigidity duration as well as the correspondent date is reported. The number of observations is different product by product. The most complete series is for blanket (1970-1995). The shortest one is for sheet (1980-1994). Per semester average change rates - reported in figure 1 - vary from a minimum of 2.15% to a maximum of 6.27%. Blanket, bags-holder and plastic boxes show the highest average change rates with the highest standard deviations. We note that for all brands we have recorded price rigidity for at least 2 semesters. The items with the smallest rigidities are pressure cooker (2 semesters), blanket and beater (3 semesters). At maximum, we observe a period of rigidity of 10 semesters for mattress-cover. The longest periods of rigidity occurred both in the low inflation regime from 1973 to 1984 and in the two-digit inflation period from 1985 to 1995 that have interested the Italian economy over the last 25 years (see figure 2). We point out that prices often were price points during these periods of rigidity. A classification of price change rates in different groups has been given as follows: <0%, =0%, 0-5%, 5-10%, 10-15%, and >15%. We have calculated frequencies for these intervals collected in table 2 and depicted in figure 3. We remark that the magnitude of change rates is variable product by product, as we have already observed in figure 1, and period by period, as we can see in table 2. In table 2 we see that the most important group is 0% for 8 goods (sheet, mattress-cover, bedspring-cover, beater, plastic boxes, bags-holder, blood pressure gauge, and electric cables); only for blanket and pressure cooker is the main frequency group 5-10% even though the 0% frequency is closed to it. We also point out the presence of discount rates whose frequency varies from 2.33% to 22.73%; only a few prices are connected to sales (special offers) which are always valid for 6 months. The 0-5% group is not negligible: 6.52% and 27.27% are minimum and maximum frequencies respectively. Data collected in table 2 were used to depict figure 2 as proposed by Kashyap (1995). Every rectangle represents a product; every area of a given color corresponds to a frequency group. For example, let's consider the mattress-cover item: discounts represent the 2.33% of the entire time series, rigidity represents the 53.49%, 0-3% frequencies are absent, 3-5% group is equal to 13.95%, 5-10% to 11.63%, 1015% to 13.95%, and variations greater than 15% amount to 4.65%. In figure 2 we can see how important rigidities are (striped area); for the 0% group we have recorded frequencies from a minimum of 18% (blanket) to a maximum of 53.49% (mattress-cover). A more precise analysis of minimum and maximum prices and of semesters during which they were chosen suggested the presence of the highest price levels in periods of inflation with two digits. Data are reported in table 3. In addition, we calculated median change rates to avoid outlyers. We 6 found that the median change rate is 0% (rigidity) for about a half of the goods. We therefore decided to consider two sub-samples; the first one is relative to catalogues n. 26-49 (spring/summer 1973 - fall/winter 1984), and the second one covers catalogues n. 50-71 (spring/summer 1985 - fall/winter 1995). Detailed statistics for both samples are presented in table 4. As we can see in table 4, at a first sight average change rates are smaller in the low inflation period than in the high inflation one. We have tested the null hypothesis of equal means in the two time periods for all items. We cannot reject it, except for blanket, bedspring cover and beater cases7. Median change rates are calculated again with reference to the two sub-samples. The 0% change rate is still the median rate for about a half of the items during both the two-digit inflation regime and low inflation one. This indicates that rigidities concern both periods. Standard deviations are high when inflation is high. Maximum change rates are recorded in the two digits inflation period, all (strictly positive) minimum change rate in the second period. Discounts in the second period are lower than discounts in the first period for 5 items. We notice however the presence of discounts in the high inflation regime too. Discounts are even higher in the latter period than in the low inflation period for 6 brands. 3. Consistency of nominal prices and state-dependent rules In a one-sided (S, s) rule model the nominal price varies only when the real price reaches the lower limit s8. In a situation of steady stochastic processes of costs, demand, and competition conditions, the target price S and the lower bound s do not change; therefore we should always observe the same real price adjustment, that is a real change of magnitude (S - s). We will check this proposition in next section. Since the nominal adjustment should be optimal only when the real price decreases of a certain amount, we could not observe small nominal variations. So it is difficult to justify the presence of them. In our data set we observe that the frequency of nominal prices in the 05% nominal change rates interval is not negligible (from 6.52% to 27.27%). See table 2. Thus, data appear to be in contrast with theoretical implications of fixed (S, s) rules. However, in a more detailed analysis of state-dependent rules, we can show that small changes are still possible. If a nonsteady stochastic process describes inflation, demand, and cost conditions, the optimal rule is to fix a 7 We cannot refuse a hypothesis of different variances in the sub-samples for two items, beater and plastic boxes. For the other products, we can accept an equal variance assumption. 8 A one-sided rule is optimal if the inflation process is monotone and stochastic (Sheshinski and Weiss 1977, 1983). If the monotonicity hypothesis cannot be satisfied, a firm optimally chooses a two-sided pricing rule (Caplin and Spulber, 1987). 7 variable band within which the real price fluctuates and the nominal price does not change9. More specifically, if the expected inflation rate increases, the bandwidth becomes larger; viceversa, if it decreases, the bandwidth gets smaller. Demand and cost shocks are very important in the determination of this interval. In the latter situation, small nominal changes are admitted. Another issue we are interested in is the (positive) correlation between the magnitude of nominal change rates and rigidity duration, as suggested by Carlton (1986). This is a useful way to check if an adjustment cost exists. For those goods whose price is very rigid we should find high opportunity costs in case of adjustment, then we should observe only large variation of nominal prices. In table 5, we observe high average change rates if the interval of rigidity is higher than one semester (n ≥ 2). In the case of maximum flexibility (the nominal price changes in every catalogue), the average change rate is low for all goods, except for sheet and blood pressure gauge. Another phenomenon we have checked, whose validity is commonly accepted in real experience, is downward rigidity. We have considered two different explanations. First, Carlton (1986) explains downward rigidity with the presence of asymmetric costs of adjustment. The cost is higher for decreases than for positive changes of nominal prices. In this case we should observe smaller minimum strictly positive change rates than minimum strictly negative change rates (in absolute value). In our sample, 6 items do not satisfy this condition (sheet, blanket, mattress-cover, bedspring-cover, beater, and plastic boxes) as we can observe in table 6; the condition holds for the other 5 goods (pressure cookers 5 and 7 lt., bags-holder, blood pressure gauge, and electric cables). In a different model by Tsiddon (1991, 1993) downward rigidity is admitted with a variable (S, s) rule and a revision of inflation expectations. As a consequence, we do not necessarily have to assume asymmetric costs of adjustment. When expectations are revised, decreases of nominal prices are less frequent than increases since it is easier that the fixed adjustment cost be lower than the opportunity cost of waiting the real price enters inside the band again. Tsiddon showed that if expected inflation rate is high, the firm chooses a large band in the (S, s) approach. If it decreases, the firm shortens the band. If the real price is very low when there is a revision like the latter, the nominal price is adjusted because the real price cannot exogenously enter the new band, independently of the adjustment cost. If the real price is very high, inflation reduces the real price bringing it spontaneously inside the new optimal band. In the latter situation, the nominal price is not adjusted if the adjustment cost is greater 9 The optimality of a variable (S, s) rule is guaranteed when the firm’s choice is related to the expected inflation rate (Tsiddon 1991, 1993); another way to follow such a rule is to assume the presence of strategic complementarities among firms (Caballero and Engel, 1993b). 8 than the opportunity cost from rigidity. This is the reason decreases are less frequent than increases. These results strictly reflect the behavior of inflation. In addition we have looked at price setting chosen by a firm. Firms often fix prices at special levels, defined in marketing theory as price points; they represent barriers the firm gets over with difficulty. We have checked if in our sample we find some evidence about price points. In figure 4 we see how much these prices (expressed in Italian lire) are important; the 54.44% of prices for all 11 items (245 prices over 450) have the hundreds which belong to the interval 900-1000. Taking into account price points, the average change rate, when a price point changes, is usually bigger than the average rate as a non-price-point varies (except for sheet, pressure cooker 7 lt., beater, plastic boxes, and electric cables; for some of them only one price belongs to the latter group). These results are reported in table 7. Finally the presence of discounts can be interpreted as expectations’ revision; if a (S, s) rule is applied adjustment is not always a convenient decision, depending on the comparison between the (fixed) adjustment cost and the opportunity-cost from rigidity. An alternative explanation of this issue can be related to the adjustment of stocks10. We often observe that firms can offer goods at low prices when they buy very big quantities. The idea however cannot find confirmation from our evidence, because we do not have data on stock dynamics. 4. Consistency of real prices and state-dependent rules Nominal changes are strongly influenced by price behavior common to all goods, summarized by the CPI change rate. To construct real prices suitable we have considered the “nominal price/general Consumer Price Index” ratio. We have calculated per semester change rates and correspondent mean, standard deviation, maximum and minimum rates; data are all summarized in table 8. Average change rates of relative prices are depicted in figure 5. We can observe that average real price change rates vary over the sample from -2.60% of electric cables to 1.41% of plastic boxes. An analysis of autocorrelations of real change rates provides evidence in favor of constant rates over time: real rate at time t is equal to a constant plus a (white noise) error term. This result is confirmed by a deeper analysis, which compares two sub-samples related to different inflation regimes observed in Italy: 1970-1984 and 1985-1995. In table 9, we reported summarizing data for both samples. An F-test has been 15 References Aguirregabiria, V. (1995) Moment Conditions in Dynamic Decision Models with Corner Solutions: Estimating a Model of Price and Inventory Decisions, University of Western Ontario, mimeo. Akerlof, George A., and Janet L. Yellen (1985) A Near-Rational Model of the Business Cycle, with Wage and Price Inertia, Quarterly Journal of Economics, Supplement, 823-838. Barro, Robert (1972) A Theory of Monopolistic Price Adjustment, Review of Economic Studies, 34, January, 17-26. Blinder, Alan S. (1991) Why Are Prices Sticky? Preliminary Results from an Interview Study, American Economic Review, Papers and Proceedings, 89-96. Caballero, R.J. e Engel, M.R.A. (1993a) Microeconomics Rigidities and Aggregate Price Dynamics. European Economic Review, 37, 697-717. Caballero, R.J. e Engel, M.R.A. (1993b) Microeconomic Adjustment Hazards and Aggregate Dynamics. Quarterly Journal of Economics, 108, 359-383. Caballero, R.J. e Engel, M.R.A. (1993c) Heterogeneity and Output Fluctuations in a Dynamic Menu-Cost Economy. Review of Economic Studies, 60, 95-120. Caplin, A. e Spulber, D.F. (1987) Menu Costs and the Neutrality of Money. Quarterly Journal of Economics, 102,703-725. Carlton, Dennis (1986) The Rigidity of Prices. American Economic Review, 637-658. Cecchetti, Stephen G. (1986) The Frequency of Price Adjustment: A Study of the Newsstand Prices of Magazines, Journal of Econometrics, 255-274. Fischer, S. (1977) Long-term Contracts, Rational Expectations, and the Optimal Money Supply Rule, Journal of Political Economy, 85, 191-205. Friedman, L. (1967) Psychological Pricing in the Food Industry, in Prices: Issues in Theory, Practice and Public Policy, A. Phillips e O.E. Williamson (eds.), Philadelphia: University of Pennsylvania Press. Gordon, Robert J. (1990) What Is New-Keynesian Economics?, Journal of Economic Literature, 1115-1171. Greene, William H. (1993) Econometric Analysis, Prentice Hall. Greenwald, Bruce e Joseph Stiglitz (1993) New and Old Keynesians, Journal of Economic Perspectives, 7, 23-44. 16 Kashyap, Anil K. (1987) Sticky Prices: New Evidence from Retail Catalogs, mimeo. Kashyap, Anil K. (1995) Sticky Prices: New Evidence from Retail Catalogs, Quarterly Journal of Economics, 245-274. Mussa (1981) Sticky Prices and Disequilibrium Adjustment in a Rational Model of the Inflationary Process, American Economic Review, 71, 1020-1027. Sheshinski, E. e Weiss, Y. (1977) Inflation and Costs of Price Adjustment. Review of Economic Studies, 44, 287-303. Sheshinski, E. e Weiss, Y. (1983) Optimum Pricing Policy under Stochastic Inflation. Review of Economic Studies, 50, 513-529. Stiglitz, Joseph (1979) Equilibrium in Product Markets with Imperfect Information, American Economic Review, 71, 393-410. Taylor, John B. (1979) Staggered Wage Setting in a Macro Model, American Economic Review, 108-113. Taylor, John B. (1980) Aggregate Dynamics and Staggered Contracts, Journal of Political Economy, 1-23. Tsiddon, D. (1991) On the Stubbornness of Sticky Prices. International Economic Review, 32, 69-75. Tsiddon, D. (1993) The (Mis)Behaviour of the Aggregate Price Level. Review of Economic Studies, 60, 889-902. Weiss, Yoram (1993) Inflation and Price Adjustment: A Survey of Findings from Micro-Data, in Optimal Pricing, Inflation, and the Cost of Price Adjustment, E. Sheshinski e Y. Weiss (eds.). Cambridge, MA: MIT Press. 17 Table 1: Descriptive statistics Item Sample Mean St. dev. Longest period of rigidity Dates Duration (sem) Sheet 80N-94N 3,28 8,49 84N-87M 6 Blanket 71M-95N 5,43 9,25 80N-81N/91M-92 3 Mattress-cover 74N-95N 4,27 6,66 86N-91M 10 Bedspring-cover 72M-95N 4,51 6,38 87N-91M 8 Pressure cooker 5 lt 77N-90N 4,30 8,21 many 2 Pressure cooker 7 lt 77N-92N 4,26 8,68 many 2 Beater 72M-95N 4,39 8,46 71N-72N 3 Plastic boxes 72M-95N 6,63 13,61 71N-73M 4 Bags-holder 72M-88M 6,14 12,74 71N-73M/77N-79M 4 Blood pressure gauge 79M-95N 3,23 8,18 86M-88N 5 Electrical cables 73M-95N 2,77 7,51 82M-84N 6 Table 2: Distribution of nominal price change rates Item < 0 = 0 0%-5% 5%-10% 10%-15% ≥≥15% Sheet 10,34% 58,62% 10,34% 3,45% 6,91% 10,34% Blanket 16% 18% 20% 20% 14% 12% Mattress-cover 2,33% 53,49% 13,95% 11,63% 13,95% 4,65% Bedspring-cover 4,35% 47,83% 6,52% 21,74% 13,04% 6,52% Pressure cooker 5 lt 9,10% 21,21% 27,27% 30,30% 6,06% 6,06% Pressure cooker 7 lt 9,10% 21,21% 27,27% 33,33% 3,03% 6,06% Beater 22,73% 22,73% 13,64% 18,17% 13,64% 9,09% Plastic boxes 5% 40% 22,50% 17,50% 5% 10% Bags-holder 7,69% 38,47% 12,82% 10,26% 15,38% 15,38% Blood pressure gauge 17,65% 38,23% 17,65% 14,71% 8,82% 2,94% Electrical cables 10,87% 47,83% 8,69% 21,74% 4,35% 6,52% 18 Table 3: Average, maximum and minimum nominal price change rates Item Mean St. dev. Max Min Median Sheet 3.28 8.49 33.44 -11.56 0.00 Blanket 5.43 9.25 33.47 -18.71 4.90 Mattress-cover 3.47 5.78 27.19 -3.39 0.00 Bedspring-cover 4.05 6.30 19.05 -15.09 0.02 Pressure cooker 5 lt 4.10 7.40 28.54 -12.02 3.75 Pressure cooker 7 lt 3.91 7.90 32.95 -14.31 2.94 Beater 3.56 7.72 24.37 -16.26 0.48 Plastic boxes 6.05 13.72 60.40 -8.11 0.07 Bags-holder 6.27 13.89 60.61 -12.58 0.00 Blood pressure gauge 3.23 8.18 41.13 -5.09 0.00 Electrical cables 2.15 6.43 21.62 -23.33 0.00 Table 4: Analysis of nominal price change rates in high-inflation and low-inflation sub-samples Sample: 1973M-1984N Sample: 1985M-1995N Item Mean St. dev. Max Min Median Mean St. dev. Max Min Median Sheet 5.51 12.66 33.44 -11.56 0.00 2.27 5.93 19.42 -3.88 0.00 Blanket 8.15 10.13 33.47 -5.27 6.05 2.12 7.48 13.07 -18.71 3.02 Mattress-cover 5.11 7.28 27.19 -3.39 0.25 1.91 3.32 11.15 0.00 0.00 Bedspring-cover 5.86 6.67 19.05 -0.03 2.82 1.68 5.35 10.37 -15.09 0.00 Pressure cooker 5 lt 3.59 9.56 28.54 -12.02 0.11 4.58 4.87 20.04 -0.09 4.09 Pressure cooker 7 lt 3.72 10.59 32.95 -14.31 0.12 4.09 4.47 18.68 -0.08 3.46 Beater 5.79 7.13 24.37 -4.14 4.63 1.45 8.14 21.36 -16.26 0.00 Plastic boxes 5.69 12.72 50.63 0.00 0.00 6.98 15.48 60.40 -8.11 2.86 Bags-holder 8.25 15.38 60.61 -12.20 0.00 1.24 8.03 13.33 -12.58 1.35 Blood pressure gauge 2.42 4.38 9.13 -4.18 0.00 3.68 9.72 41.13 -5.09 0.00 Electrical cables 2.46 8.11 21.62 -23.33 0.00 1.82 4.04 10.10 -5.29 0.00 19 Table 5: Average change rate (in absolute value) after a period of rigidity of n semesters Item n=1 n>1 weighted average n>1 simple average n=2 n=3 n=4 n=5 n=6 n=8 n=10 Sheet 11.19 9.97 7.90 14.62 /7.98 -1.09 / / / Blanket 8.62 11.16 8.73 9.01 16.56 / / / / / Mattress-cover 7.69 9.115 10.69 12.30 /27.19 10.03 11.15 /3.86 Bedspring-cover 8.89 10.93 12.68 8.75 18.52 19.05 /6.71 10.37 / Pressure cooker 5 lt. 6.59 13.73 13.73 13.73 / / / / / / Pressure cooker 7 lt. 6.21 13.36 13.36 13.36 / / / / / / Beater 6.95 10.70 8.79 11.34 6.25 / / / / / Plastic boxes 7.33 17.19 19.88 30.68 9.09 / / / / / Bags-holder 10.29 19.22 16.28 25.09 12.92 10.83 / / / / Blood pressure gauge 8.96 5.00 4.77 4.76 6.67 /2.87 / / / Electrical cables 7.15 10.48 8.24 5.06 13.35 9.17 /5.37 / / Table 6: Downward rigidity Item Strictly positive minimum change rate Strictly negative minimum change rate Sheet 2.11 -1.09 Blanket 1.70 -0.19 Mattress-cover 3.86 -3.39* Bedspring-cover 4.32 -2.86 Pressure cooker 5 lt. 0.13 -7.42 Pressure cooker 7 lt. 0.12 -6.68 Beater 0.72 -0.52 Plastic boxes 0.85 -0.84 Bags-holder 1.35 -2.16 Blood pressure gauge 2.12 -2.79 Electrical cables 1.27 -2.23 * This is the only negative change rate in the mattress-cover price series. 20 Table 7: Average nominal price change rates and price points Item Average change rate after a price point is adjusted (n° of changes) Average change rate after another price is adjusted (n° of changes) Number of semesters with a 0% change rate Overall average change rate (n° of changes) Sheet 6.90% (11) 19.42% (1) 17 3.28% (29) Blanket 8.40% (30) 1.78% (11) 95.43% (50) Mattress-cover 9.12% (18) 6.81% (2) 23 3.55% (43) Bedspring-cover 8.63% (18) 6.75% (6) 22 4.26% (46) Pressure cooker 5 lt. 6.25% (21) 2.72% (5) 74.39% (33) Pressure cooker 7 lt. 5.33% (23) 5.81% (3) 74.24% (33) Beater 4.75% (29) 6.54% (5) 10 3.87% (44) Plastic boxes 11.20% (22) 21.71% (2) 16 7.24% (40) Bags-holder 11.73% (18) 10.58% (6) 15 7.04% (39) Blood pressure gauge 5.23% (21) /13 3.23% (34) Electrical cables 5.22% (23) 7.25% (1) 22 2.77% (46) 21 Table 8: Mean, maximum, and minimum change rates of real prices Item Mean St. Dev. Max Min Sheet -0,85 7,40 21,25 -14,91 Blanket 0,24 8,31 25,97 -20,78 Mattress-cover -0,99 6,00 18,12 -14,24 Bedspring-cover -0,98 5,22 14,55 -17,34 Pressure cooker 5 lt. -0,71 7,37 15,98 -16,72 Pressure cooker 7 lt. -0,76 7,37 19,95 -18,88 Beater -0,67 7,50 18,12 -18,72 Plastic boxes 0,16 11,20 40,92 -12,77 Bags-holder -0,12 12,69 54,53 -17,01 Blood pressure gauge -1,04 8,77 38,28 -12,94 Electrical cables -2,60 7,02 15,94 -26,96 Table 9: Real price change rates in two different inflation regimes Two-digit inflation (1973 - 1983) Low inflation (1984 - 1995) Item Mean St.Dev. Max Min Mean St.Dev. Max Min Sheet 0,60 10,21 21,25 -8,85 -1,31 6,50 16,23 -14,91 Blanket 0,71 9,65 25,97 -10,62 -0,25 6,84 10,78 -20,78 Mattress-cover -1,02 8,22 18,12 -14,24 -0,97 3,29 9,08 -4,80 Bedspring-cover -1,17 4,39 6,88 -7,89 -0,79 6,07 14,55 -17,34 Pressure cooker 5 lt. -2,51 8,96 15,98 -16,72 0,59 5,90 13,96 -11,96 Pressure cooker 7 lt. -2,26 10,02 19,95 -18,88 0,33 5,44 12,66 -11,00 Beater -0,45 7,09 12,15 -11,23 -0,91 8,06 18,12 -18,72 Plastic boxes -2,50 9,86 38,06 -11,23 3.07 12.08 40.92 -12,77 Bags-holder -1,84 8,14 18,61 -10,07 4,29 20,29 54,53 -17,01 Blood pressure gauge -6,03 4,19 0,83 -12,94 1,23 9,42 38,28 -6,99 Electrical cables -3,71 9,17 15,94 -26,96 -1,49 3,76 3,86 -8,07 22 Table 10: Average real loss when nominal prices are sticky for at least 2 semesters Item Average real loss Sheet 9.47% Blanket 6.87% Mattress-cover 16.08% Bedspring-cover 9.49% Pressure cooker 5 lt. 6.54% Beater 6.85% Plastic boxes 11.13% Bags-holder 11.74% Blood pressure gauge 7.65% Electrical cables 12.42% 23 Table 11: Ordered-probit model Explanatory variables Coefficients Std. Err. zP>|z| [95% Conf. Interval] realp1 -.0058514 .0032105 -1.823 0.068 -.0121438 .0004411 CPIrate .0382261 .0154026 2.482 0.013 .0080376 .0684147 d1 -2.146494 .7756174 -2.767 0.006 -3.666676 -.6263121 d2 -.4625284 .3043673 -1.520 0.129 -1.059077 .1340204 d3 -1.861539 .7737311 -2.406 0.016 -3.378024 -.3450539 d4 -1.814591 .7850343 -2.311 0.021 -3.35323 -.275952 d5 -.1810895 .3434961 -0.527 0.598 -.8543296 .4921505 d7 -1.61791 .6737057 -2.402 0.016 -2.938349 -.2974713 d8 -1.69395 .7829515 -2.164 0.030 -3.228507 -.1593931 d9 -2.125875 .8415598 -2.526 0.012 -3.775302 -.4764479 d10 -1.359852 .445535 -3.052 0.002 -2.233084 -.4866191 d11 -2.118553 .7881334 -2.688 0.007 -3.663266 -.5738403 k1-3.163398 .8517366 k2-1.881343 .8447463 Table 12: Results in single product probit models CPI change rate (t -1) Real price (t -1) Item Coefficient St.Dev. P-value Coefficient St.Dev. P-value Sheet 0.0637095 0.1248580 0.610 0.0059477 0.0300858 0.843 Blanket 0.2116191 0.0799648 0.008 -0.0159122 0.0058820 0.007 Mattress-cover -0.0186059 0.0607129 0.759 0.0536038 0.0434952 0.218 Bedspring-cover 0.1244833 0.0662254 0.060 -0.1366657 0.0592385 0.021 Pressure cooker 5 lt. -0.0863065 0.1133675 0.446 -0.0175080 0.0119326 0.142 Beater 0.0936233 0.0592224 0.114 -0.0249542 0.0242500 0.303 Plastic boxes 0.0506703 0.0513897 0.324 0.0247338 0.0202136 0.221 Bags-holder 0.0098132 0.0514722 0.849 -0.0998430 0.0833861 0.231 Blood pressure gauge 0.0420656 0.0643689 0.513 -0.0115078 0.0098953 0.245 Electrical cables -0.0035274 0.0311621 0.910 -0.0205131 0.0411381 0.618 24 Table 13: Predicted probabilities in the pooled estimation Pooling model -1 0+1 Predicted probability 0.1016 0.3887 0.5097 Relative frequency 0.1059 0.3892 0.5049 Table 14: Predicted probabilities in single-product estimates Predicted prob. Actual frequency Item -1 0+1 -1 0+1 Sheet 0.1010 0.5881 0.3109 0.1034 0.5862 0.3103 Blanket 0.1519 0.1747 0.6734 0.16 0.18 0.66 Mattress-cover 0.0252 0.5350 0.4398 0.0233 0.5349 0.4419 Bedspring-cover 0.0485 0.4438 0.5077 0.0417 0.4583 0.5 Pressure cooker 0.0937 0.1834 0.7229 0.1111 0.2222 0.6667 Beater 0.2214 0.2035 0.5751 0.2292 0.2083 0.5625 Plastic boxes 0.0437 0.3903 0.5659 0.0417 0.3958 0.5625 Bags-holder 0.0559 0.5258 0.4183 0.0606 0.5152 0.4242 Blood press. gauge 0.1469 0.3989 0.4542 0.1765 0.3824 0.4412 Electrical cables 0.1138 0.4769 0.4093 0.1087 0.4783 0.4130